gis-based spatial model for wildfire simulation:...
TRANSCRIPT
GIS-BASED SPATIAL MODEL FOR WILDFIRE SIMULATION: MARMARİS – ÇETİBELİ FIRE
A THESIS SUBMITED TO THE GRADUATE SCHOOL OF NATURAL AND APPLIED SCIENCES
OF THE MIDDLE EAST TECHNICAL UNIVERSITY
BY
ERDİNÇ TAŞEL
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE
IN THE DEPARTMENT OF
GEODETIC AND GEOGRAPHIC INFORMATION TECHNOLOGIES
NOVEMBER 2003
Approval of the Graduate School of Natural and Applied Sciences.
Prof. Dr. Canan Özgen
Director
I certifty that this thesis satisfies all the requirements as a thesis for the degree of
Master of Science.
Assoc. Prof. Dr. Oğuz Işık
Head of Department
This is to certify that we have read this thesis and that in our opinion it is fully
adequate, in scope and quality, as a thesis for degree of Master of Science.
Assist. Prof. Dr. Zuhal Akyürek
Supervisor
Examining Committee Members
Assoc. Prof. Dr. Nurünnisa Usul
Assoc. Prof. Dr. Vedat Toprak
Assist. Prof. Dr. Zuhal Akyürek
Assist. Prof. Dr. Lütfi Süzen
Y. Müh. Yüksel Erdoğan
iii
ABSTRACT
GIS-BASED SPATIAL MODEL FOR WILDFIRE
SIMULATION: MARMARİS – ÇETİBELİ FIRE
TAŞEL, ERDİNÇ
M.S., Department of Geodetic and Geographic Information Technologies
Supervisor: Assist. Prof. Dr. Zuhal Akyürek
November 2003, 175 Pages
Each year many forest fires have occurred and huge amount of forest areas
in each country have been lost. Turkey like many world countries have forest fire
problem. 27 % of Turkey’s lands are covered by forest and 48 % of these forest
areas are productive, however 52 % of them must be protected. There occurred
21000 forest fires due to several reasons between 1993 and 2002. It is estimated
that 23477 ha area has been destroyed annually due to wildfires.
The fire management strategies can be built on the scenarios derived from
the simulation processes. In this study a GIS – based fire simulating model is used
to simulate a past fire occurred in Marmaris – Çetibeli, Turkey, in August 2002.
iv
This model uses Rothermel’s surface fire model, Rothermel’s and Van Wagner’s
crown fire model and Albini’s torching tree model. The input variables required
by the model can be divided into four groups: fuel type, fuel moisture, topography
and wind. The suitable fuel type classification of the vegetation of the study area
has been performed according to the Northern Forest Fire Laboratory (NFFL)
Fuel Model. The fuel moisture data were obtained from the experts working in the
General Directorate of Forestry. The fire spread pattern was derived using two
IKONOS images representing the pre- and post-fire situations by visual
interpretation. Time of arrival, the rate of spread and the spread direction of the
fire were obtained as the output and 70 % of the burned area was estimated
correctly from the fire simulating model.
Keywords: Fire Management, Fire Behavior Model, GIS, NFFL Fuel Model,
Marmaris – Çetibeli.
v
ÖZ
CBS TABANLI MEKANSAL ORMAN YANGIN
SİMULASYONU: MARMARİS ÇETİBELİ YANGINI
TAŞEL, ERDİNÇ
Yüksek Lisans, Jeodezi ve Coğrafi Bilgi Teknolojileri Bölümü
Tez Yöneticisi: Y. Doç. Dr. Zuhal AKYÜREK
Kasım 2003, 175 Sayfa
Diğer dünya ülkelerinde olduğu gibi Türkiye’de de orman yangını problem
teşkil etmektedir. Her yıl çok sayıda orman yangını meydana gelmekte ve bu
yangınlar sonucunda çok geniş orman alanları yok olmaktadır. Türkiye
topraklarının % 27’si ormanlarla örtülü olup bu alanın % 48’ si ekonomik açıdan
değerlidir, fakat bu ormanlık alanların % 52’ si koruma altındadır. 1993 ile 2002
yılları arasında yaklaşık 21000 adet orman yangını çeşitli nedenlerden ötürü
meydana gelmiş, yıllık kaybedilen orman alanı 23477 hektar olarak belirlenmiştir.
Orman yangını yönetimi, simülasyonlar sonucunda elde edilen orman
yangını senaryoları üzerine kurulabilir. Bu çalışmada Coğrafi Bilgi Sistemleri
vi
(CBS) tabanlı yangın simülasyon modeli kullanılarak 2002 Ağustos ayında
Marmaris – Çetibeli’nde meydana gelen orman yangını simüle edilmiştir. Bu
model Rothermel örtü yangın modeli, Van Wagner tepe yangın modeli ve Albini
tutuşan ağaç modellerıni bileşik bir şekilde kullanmaktadır. Modelin çalışması
için gerekli veri değişkenleri yakıt cinsi, yakıt nemi, topoğrafya ve rüzgar olarak 4
gruba ayrılır. Çalışma alanının yakıt modeli Northern Forest Fire Laboratory
(NFFL) yakıt modeline göre en uygun olacak şekilde belirlenmiştir. Yakıt nem
verisi Orman Genel Müdürlüğü’nde çalışan uzmanlardan elde edilmiş olup yangın
öncesi ve sonrasına ait IKONOS görüntüleri kullanılarak yangının yayılım dokusu
görsel olarak belirlenmiştir. Yangın simülasyon modeli tarafından yangının
ulaşım zamanı, yayılım hızı ve yönü çıktı olarak oluşturulmuş ve toplam yanan
alanın % 70’ i doğru olarak tahmin edilmiştir.
Anahtar Kelimeler: Yangın Yönetimi, Yangın Davranış Modeli, CBS, NFFL
Yakıt Modeli, Marmaris – Çetibeli.
vii
Dedicated to people who sacrificed their lives for Green Earth
viii
ACKNOWLEDGEMENTS
I wish to thank the following people and association for their support, help
and advices during the preparation of this thesis.
My supervisor, Assist. Prof. Dr. Zuhal Akyürek for her kind supervision,
support and contribution throughout this research.
The staff of General Directorate of Forest, especially Fire Operation
Center Forest Engineers, for their help and providing information.
Mr Seyit Demirci, Map Fiduciary of METU, for providing the topographic
maps of the study area.
Mr. Okan Turan, Çetibeli Forest Region Chief, for providing information
about Çetibeli forest fire.
United States Department of Agriculture (USDA) Forest Service for
sending invaluable journals.
INTA Space Imaging Eurasia for providing satellite images of the study
area.
Thank to my family, mom, dad and brother, for physical and moral full
support during thesis.
Elif Yurdanur, my love, for her support and helps.
All the people, whose names have not been written here, for their help.
ix
TABLE OF CONTENTS
ABSTRACT …………………………………………………………………
ÖZ ....................................................................................................................
DEDICATION ……………………………………………………………….
ACKNOWLEDGMENTS …………………………………………………...
TABLE OF CONTENTS ……………………………………………………
LIST OF TABLES …………………………………………………………...
LIST OF FIGURES ………………………………………………………….
CHAPTER
1. INTRODUCTION ……………………………………………….
1.1. Objectives …………………………………………………...
1.2. Outline of the Thesis ………………………………………...
2. FOREST FIRES ………………………………………………….
2.1. What are Forest Fires? …………………………………...….
2.2. Character of Forest Fires …………..………………………..
2.3. The Impacts of Forest Fires …………………...…….............
2.4. The Origins of Forest Fires in Turkey ...……………….........
2.5. Forest Fires in Turkey ……………………………………….
2.6. Using Geographic Information Systems in Fire Management
3. FARSITE: FIRE AREA SIMULATOR MODEL ……………….
3.1. The Application Areas ………………………………………
3.2. Capabilities ………………………………………………….
3.3. Data Requirements .………………………………………....
3.3.1. Landscape File ……………………………………….
3.3.2. ASCII Input Files …………………………………….
iii
v
vii
viii
ix
xi
xii
1
4
5
6
6
10
15
18
20
25
34
35
38
40
41
50
x
3.4. Limitations and Assumptions ……………………………….
4. THE CASE STUDY: MARMARİS – ÇETİBELİ WILDFIRE …
4.1. The Study Area ..…….………………………………………
4.2. The GIS Data Preparation …………………………………...
4.3. Generation of Landscape Input File ..……………………….
4.4. Creating ASCII Input Files ...………………………………..
4.5. Simulation Process of Çetibeli Wildfire ...…………………..
4.5.1. Çetibeli Fire Simulation Run ..……………………….
4.5.2. Calibration and Adjustment of Çetibeli Fire
Simulation ……………………………………………...
4.6. Discussion of the Results …...……………………………….
5. CONCLUSIONS AND RECOMMENDATIONS ........................
5.1. Recommendation for Future Work ………………………….
REFERENCES ................................................................................................
APPENDICES
A. NFFL FUEL MODEL DESCRIPTION …………………..
B. PARAMETERS AND EXAMPLES OF FARSITE ASCII
INPUT FILES ……………………………………………..
C. FIRE REGISTER REPORT ………………………………
D. LINEAGE REPORT ……………………………………...
E. TABLE OF VEGETATION SYMBOLS ………………...
F. RAW METEOROLOGICAL DATA ……………………..
G. ASCII INPUT FILES OF ÇETİBELİ .................................
H. SIMULATION LOG FILES ......………………………….
I. FIRE OBSERVATION DATA SHEETS AND
DESCRIPTIONS ……………………………………………
56
60
61
66
72
76
77
90
93
98
104
107
109
113
140
148
150
156
158
162
166
171
xi
LIST OF TABLES
TABLE
2.1 The Number and Burned Area of Forest Fires in Some
Mediterranean Countries ………………………………………
8
2.2 The Situation of the Turkey’s Forests ………………………… 11
2.3 The Classes of Fires According to Sizes ……………………… 23
3.1 FARSITE Input Data Files ……………………………………. 42
3.2 Raster File Data Units …………………………………………. 48
3.3 Raster inputs to FARSITE and their usage in the simulation …. 49
4.1 The Largest Forest Fires in Turkey …………………………… 62
4.2 Meteorological Conditions during Fire (obtained from Muğla
Meteorological Station) ………………………………………..
64
4.3 Total Cost of Çetibeli Fire .......................................................... 66
4.4 Raster Output Themes Filename Extensions and Units ………. 89
D.1 Comparison of 10 Elevation Data Which Were Read From
DEM and Topographic Map …………………………………...
154
D.2 Lineage Information of Digital Maps …………………………. 155
E.1 Table of Vegetation Symbols Use by General Directorate of
Forest …………………………………………………………..
156
xii
LIST OF FIGURES
FIGURE
2.1 Fire Triangle ………………………………………………….. 7
2.2 The Number of Forest Fires in Some Mediterranean Countries 9
2.3 The Burned Area of Forest Fires in Some Mediterranean
Countries ……………………………………………………...
9
2.4 Types of Fires ………………………………………………… 10
2.5 The Reasons of the Forest Fires in Turkey in 2002 ………….. 19
2.6 The Overall View of the Forest Areas in Turkey …………….. 21
2.7 The Distribution of Number of Forest Fires between 1993-
2002 …………………………………………………………...
21
2.8 The Distribution of Burned Area between 1993-2002 ……….. 22
2.9 The Head Offices of General Directorate of Forest in Turkey . 24
2.10 The Distribution of Forest Areas According to Fires Risk
Degrees in 2002 ………………………..……………………..
26
2.11 The Distribution of Red Pine Forest Areas According to Fire
Risk Degrees in 2002 …………………………………………
27
2.12 The Distribution of Forest Fires According to Number of Fire
in 2002 ………………………………………………………...
28
2.13 The Distribution of Forest Fires According to Area Burned in
2002 ………………………………………………………...
29
3.1 Raster Input Layers Generated by the Geographic Information
Systems for FARSITE Simulation ……...…………………….
36
3.2 Definition of Crown Fuel Data Used in FARSITE 46
4.1 Location of the Study Area …………………………………... 63
xiii
4.2 Ground Control Points on the Topographic Maps Belonging
to the Study Area ……………………………………………...
68
4.3 Pre-fire IKONOS Satellite Image, (22nd of April, 2002 at
09:05) (A) Visible Spectrum, (B) Near Infrared ……………...
69
4.4 Post-fire IKONOS Satellite Image, (21st of August, 2002 at
09:15) (A) Visible Spectrum, (B) Near Infrared ………...........
70
4.5 Data Themes of Landscape File. (A) Elevation, (B) Slope, and
(C) Aspect of Çetibeli ...............................................................
74
4.6 The (A) Fuel Model and (B) Canopy Cover of the Study Area 75
4.7 The View of (A) Ignition Point and (B) Burned Area ……….. 78
4.8 Flow Diagram of FARSITE ………………………...………... 80
4.9 Landscape (.LCP) File Generation …………………………… 81
4.10 FARSITE Project File Generation …………………………… 82
4.11 Model Parameters ……………………………………………. 83
4.12 Simulation Duration ………………………………………….. 86
4.13 Fire Behavior Options ………………………………………... 87
4.14 Export and Output Options of Çetibeli Fire Simulation ........... 90
4.15 The View of FARSITE Before the Run ……………………… 91
4.16 The View of FARSITE During the Run ……………………... 92
4.17 Simulated and Real Burned Area (First Run) ………………... 93
4.18 Fire Behavior Outputs. (A)Time of Arrival, (B) Fireline
Intensity, (C) Flame Length, (D) Rate of Spread (E) Heat Per
Unit Area, (F) Spread Direction, (G) Air Temperature ………
94
4.19 Simulated and Real Burned Area (Second Run) ……………... 96
4.20 Simulated and Real Burned Area (Final Run) ……………….. 97
4.21 Simulated and Real Burned Area (Final Run after
Modification) …………………………………………………
98
4.22 The Outputs of (A) Time of Arrival, (B) Rate of Spread and
(C) Spread Direction ………………………………………….
99
4.23 The Outputs of (D) Relative Humidity, (E) Air Temperature
and (F) Solar Radiation ……………………………………….
100
xiv
4.24 Time Contour of Fire Spread (30 min interval) and Fuel
Models ………………………………………………………..
103
1
CHAPTER 1
INTRODUCTION
Forests are one of the most valuable natural resources because of adjusting
the natural balance, affecting the climate and water body of the region, preventing
air pollution and erosion. In addition to this, they are important for community to
meet the demand of products made of timber.
Forest protection is an important part of silviculture, which is the science,
art and practice of caring for forests with respect to human objectives. There are
lots of injurious activities and many of them are so destructive. Valuable and
healthy timber production is unattainable if adequate protection and preservation
are not afforded.
Injurious agencies may be potentially very hazardous to forests and in
contrast because of silvicultural operations in woodlands, impairment may be
minimized. Injurious agencies can be listed as follows:
1. Forest fires.
2
2. Plants, including fungi, mistletoes, and forest weeds.
3. Insects.
4. Domestic animals.
5. Wildlife (animals other than insects and domestic species).
6. Atmospheric agencies. Under this heading the injuries from heat, frost,
drought, water (including injury from floods, erosion caused by water,
landslides, snow, ice, hail, and avalanches), air movements (including
the effects of drifting sand and erosion caused by wind), lightning,
poisonous gases, and smoke are included (Hawley and Stickel, 1948).
Forest fires occur either because of anthropological or natural causes. The
majority of fires around the globe are caused by human activities. Lightning is
probably the most common natural cause of fire. They have an instantaneous,
often within a few hours, and enormous destruction. They consume forests,
buildings and also cause damage to human life. The impacts of forest fires can
have global consequences: Forest fires also produce gaseous and particle emissions
that impact the composition and functioning of the jet stream and the global
atmosphere, exacerbating climate change. Tropical forest destruction, through fire,
could also spiral our weather systems in new and unpredictable directions.
Therefore, forest fires have taken the first place among injurious agencies.
According to the World Wildlife Fund (WWF), the forest areas of the
world have high wildfire risk. The Mediterranean Countries have forest fire
problem due to their location and meteorological conditions. Turkey has also large
3
amount of forest, which are extremely sensitive to fire, and they are located in
west and south regions.
The suitable response to forest fires depends on the evaluation of risks,
hazards and values, which form fire management strategies. Risk is the chance of
a fire starting. If the risk is high, fire prevention and detection are very cost-
effective. Hazard is defined as simply the amount, condition and structure of fuels
that will burn. Lastly, values are the change detection in resource condition. Fire
management requires an understanding of how fire starts and spreads; the
behavior of fires, fuels and how they are suppressed. Many fire managers are
searching the appropriate ways to manage fires rather than simply suppress them
(Edmonds et al., 2000).
Geographic Information System (GIS) can be used to model the fire
behavior where spatial and nonspatial features can be handled within the system.
Fire prediction systems model the fire behavior using site specific data, which are
weather, topography, fuel type and condition. All the needed data are spatial.
Geographic Information System technology has been gaining reputation for its
ability to integrate large amount and type of information about environmental and
public factors, which are spatially and temporally dynamic. It is preferable to
integrate fire behavior modeling into GIS framework in order to analyze the
simulation of the fire.
4
In Turkey, the responsibility of the interference to the wildfire belongs to
General Directorate of Forest in fire management. The availability of GIS – based
fire prediction models can help forest managers during struggling wildfires and
fire management. In this study, a GIS – based fire simulation model is used in a
case study of Marmaris Çetibeli fire by using FARSITE, Fire Area Simulator
Model (Finney, 1998) in order to check the suitability of the model. The
vegetation of the area was adapted to the Northern Forest Fire Laboratory (NFFL)
Fuel Model (Anderson, 1982). After that, the Çetibeli fire has been simulated in
order to test the suitability of the model and to define techniques that help to
improve the simulation.
1.1.Objectives
The main objective in this thesis is to test the suitability of a GIS – based
forest fire simulation model and determine the requirements of the model for
Turkey.
In this study, it is intended to extract the burned parts of the study area
using remote sensing techniques and ancillary data. Generation of the necessary
and suitable spatial and nonspatial input files in order to perform the simulation
and defining the vegetation types of the study area according to the NFFL Fuel
Model (Anderson, 1982) are also considered. To calibrate the parameters of the
model in order to acquire the real situation of the fire area and to find the accuracy
5
and reliability of the model by comparing the extracted and simulated burned area
in Marmaris – Çetibeli fire are additional objectives in this study.
1.2. Outline of the Thesis
In Chapter 2, in order to meet the objectives stated in Section 1.1., firstly
the character, origins and impact of forest fires were described, the forest fire
situations in Turkey and a detailed review of the literature associated with fire
behavior models, the simulation modeling systems with GIS were depicted.
In Chapter 3, application areas of a GIS – based fire simulation model,
FARSITE Fire Area Simulator and its capabilities were described. After that data
requirements and their explanations were declared. At the end of this chapter,
limitations and assumptions of fire simulation model were mentioned.
Chapter 4 gives information about Marmaris – Çetibeli fire, the study area
and the generation of input files. The simulation of the case study Marmaris –
Çetibeli wildfire were explained in detail with discussion of their results.
Chapter 5 constitutes the conclusion reached in this study and the
recommendation for future work.
6
CHAPTER 2
FOREST FIRES
Forest fires are the major ecological agent in Turkey’s forests. They can be
thought as a force of renewal in the forest, it can also place people’s lives and
property at risk. They may be seen as a natural and hence desirable process, or
they may be thought as a destroyer of forest recourse. These competing realities of
the role of fire result from the different values; whether it is seen as simply a
source of economically valuable material, or integral part of the earth’s life
support system.
2.1.What are Forest Fires?
The definition of the forest fire mainly depends on how the forest is
valued. According to industrial perspective, a forest fire consumes valuable timber
and it may be a threat for property and settlement areas even life. From tourism
and recreational perspectives forest fire forces tourist areas and destroys view
sheds. Conversely, environmental perspectives include that forest fire endangers
sensitive old-growth forest habitat and it retains forest-opening environments.
7
A natural forest fire has one of usual origins, for example lightning, and is
not caused by humans; however wildfire is an unplanned, out of control forest or
grassland fires. According to The American Heritage® Dictionary of the English
Language, Wildfire means that “A raging, rapidly spreading fire”
Fire has three components; fuel, heat, oxygen which are named as “fire
triangle”. Figure 2.1 shows the two dimensional triangle explaining the
combustion process. When all sides of the triangle are intact in proper state and
proportion, fire takes place.
Figure 2.1. Fire Triangle
The world countries have forest fire problems. The fire situation of some
Mediterranean Countries for last 10 years can be seen in Table 2.1 and graphs
were drawn in Figure 2.2 and 2.3. It has been estimated by the World Wildlife
Fund (WWF) that annually up to 500 million hectares of woodland, open forests,
tropical and sub-tropical savannahs, 10-15 million hectares of boreal and
temperate forest and 20-40 million hectares of tropical forests in the world are
burned.
HEAT FUEL
OXYGEN
FIRE
8
Area (ha)
15955
14253
19263
25828
16771
22319
22445
18237
24312
19631
199014
SPAIN
Number
105277
89331
437635
143468
59814
98503
133643
82217
188586
66075
1404549
Area (ha)
14641
15380
11588
7378
9093
11612
10155
7235
10629
7134
104845
ITALY
Number
105695
209314
68828
46466
57986
103015
140432
61989
114648
76427
984800
Area (ha)
2582
2406
1763
1438
1508
2273
1842
1486
2581
2535
20414
GREECE
Number
71410
54049
57908
27202
25310
52373
92901
8289
145033
18221
552696
Area (ha)
4008
4765
4633
6545
6400
8000
5600
5170
5600
4103
54824
FRANCE
Number
16607
16695
25872
18118
11210
20500
20880
17605
23700
17000
188187
Area (ha)
2117
2545
3239
1770
1645
1339
1932
2075
2353
2631
21646
TURKEY
Number
12232
15393
38128
7676
14922
6316
6764
5804
26353
7394
140982
Table 2.1. The Number and Burned Area of Forest Fires in Some Mediterranean Countries (General Directorate of Forests in
Turkey, 2002)
Year
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
TOTAL
9
Number of Fires
0
2500
5000
7500
10000
12500
15000
17500
20000
22500
25000
27500
Turkey France Greece Italy Spain
1992199319941995199619971998199920002001
Figure 2.2. The Number of Forest Fires in Some Mediterranean Countries
(General Directorate of Forests in Turkey, 2002)
Burned Area of Fires
0
50000
100000
150000
200000
250000
300000
350000
400000
450000
Turkey France Greece Italy Spain
Are
a (h
a)
1992199319941995199619971998199920002001
Figure 2.3. The Burned Area of Forest Fires in Some Mediterranean Countries
(General Directorate of Forests in Turkey, 2002)
10
Turkey has a significant amount of forest land areas most of them are
located through the coast of Mediterranean, Aegean Region and the coast of Black
Sea Region. The land of Turkey has been divided into regions by General
Directorate of Forest. The forests located in these regions are handled by the
directorates belonging to these regions. The information about the forests for each
region is tabulated in Table 2.2. Most of the forest areas consist of red pine
forests, which are extremely sensitive to forest fire.
2.2. Character of Forest Fires
Generally, forest fires can be classified according to the mode of spread,
type of damage and development characteristics of each kind of fires. Forest fires
are divided into 3 classes, namely ground fire, surface fire and crown fire.
Additionally, spotting is also an important event occurring during fire, as seen in
Figure 2.4.
Figure 2.4. Types of Fires (http://www.noaa.gov)
11
Red Pine Area
Class 5
Forest Area
Red Pine Area
2470
5427
12567
Class 4
Forest Area
113499
69753
373552
286847
298657
64804
Red Pine Area
10812
45789
3719
3829
Class 3
Forest Area
141170
21852
916752
274127
130037
284611
103159
Red Pine Area
87277
752
74580
60096
22946
9189
9099
28547
172076
Class 2
Forest Area
234183
86000
555591
301513
98489
242310
114871
252620
402796
Red Pine Area
193487
24085
483780
124516
97862
177082
110872
Fire Sensitivity Classes
Class 1
Forest Area
359076
180600
45201
817851
553581
506936
384127
311218
Red Pine Area
294046
24837
125796
16286
543876
147462
9189
110790
205629
282948
TOTAL
Forest Area
847927
358205
1845895
606215
1119364
428694
652070
591725
754966
636747
714014
Table 2.2. The Situation of the Turkey’s Forests (General Directorate of Forests in Turkey, 2002)
Directorate of Regions
ADANA
ADAPAZARI
AMASYA
ANKARA
ANTALYA
ARTVİN
BALIKESİR
BOLU
BURSA
ÇANAKKALE
DENİZLİ
Class 1: Most Sensitive to Forest Fire, Class 5: Less Sensitive to Forest Fire. Area Units are hectares.
12
Red Pine Area
Class 5
Forest Area
478490
117517
189031
Red Pine Area
1349
425
Class 4
Forest Area
1045402
283489
78110
77235
121103
Red Pine Area
1978
19211
112
57015
3048
Class 3
Forest Area
556197
163093
156973
82040
272484
256461
Red Pine Area
2483
52643
910
52071
2494
Class 2
Forest Area
355200
527422
262340
252356
398989
Red Pine Area
76954
19096
620029
187889
Fire Sensitivity Classes
Class 1
Forest Area
189545
170423
991255
412480
Red Pine Area
3327
21694
129597
20118
620029
296975
5967
TOTAL
Forest Area
2080089
401006
518293
424114
716967
592038
991255
937320
772553
Table 2.2. The Situation of the Turkey’s Forests (General Directorate of Forests in Turkey, 2002) (Continued)
Directorate of Regions
ELAZIĞ
ERZURUM
ESKİŞEHİR
GİRESUN
ISPARTA
İSTANBUL
İZMİR
K.MARAŞ
KASTAMONU
Class 1: Most Sensitive to Forest Fire, Class 5: Less Sensitive to Forest Fire. Area Units are hectares.
13
Red Pine Area
Class 5
Forest Area
304109
1089147
Red Pine Area
11361
33599
Class 4
Forest Area
142795
80301
39598
3075145
Red Pine Area
35078
1926
182517
Class 3
Forest Area
409198
23332
185105
434556
4411147
Red Pine Area
7901
37050
9763
629877
Class 2
Forest Area
181052
183797
239908
130743
4820180
Red Pine Area
37053
415675
726252
3294632
Fire Sensitivity Classes
Class 1
Forest Area
356307
803987
1130344
7212931
Red Pine Area
46439
44954
415675
726252
37050
11689
4140625
TOTAL
Forest Area
733045
563436
803987
1130344
320209
528812
565299
20763248
Table 2.2. The Situation of the Turkey’s Forests (General Directorate of Forests in Turkey, 2002) (Continued)
Directorate of Regions
KONYA
KÜTAHYA
MERSİN
MUĞLA
SİNOP
TRABZON
ZONGULDAK
TOTAL
Class 1: Most Sensitive to Forest Fire, Class 5: Less Sensitive to Forest Fire. Area Units are hectares.
14
“Ground fires burn in the duff, humus, and peaty layers lying beneath the
litter or undecomposed portion of the forest floor.” (Hawley and Stickel, 1948).
Ground fires burn in natural litter, duff, roots or sometimes high organic soils very
slowly but continuously with powerful heat and they are uniformly destructive
(Hawley and Stickel, 1948). Once they start, detection and control of ground fires
are very difficult.
Surface fires are the most common kind of fire that occurs on the ground
and burn the land vegetation covers, brushes and the lower branches of the trees.
This kind of fire is strongly affected by surface winds hence surface fires spread
rapidly and they have high flame height and heat, however they are deflated soon.
Usually, all fires start as surface fires. After requisite conditions occur, crown fire
may develop from surface fires.
Crown fires simply burn at the top of the trees and they depend on species
that have inflammable foliage. The conifer forests are mainly affected by crown
fires however, the foliage of most deciduous species is less flammable. There are
two types of crown fires, namely, “the running crown fire and the dependent
crown fire” (Hawley and Stickel, 1948). The running crown fire is burning
independently through the crowns of the trees and it spreads very rapidly but not
as fast as surface fires. The next one, dependent crown fire, is supported by the
surface fires and burns together. “The burning material on the ground furnishes
the volume of heat which ignites the crowns and maintains the crown fire.”
(Hawley and Stickel, 1948). When the crown fire starts, it is very difficult to take
15
under control and be suppressed since wind plays an important role in spreading
crown fires.
Spotting is a form of mass transfer, movement of heat by active firebrands,
which are lifted by convective updraft, and it is brought over the main perimeter
of the fire. Spotting distance may be extremely long. For example, in the 1991
Oakland fire in USA, the spot passed easily over eight-lane freeways. In very
intense fires with great uplifting in Washington and Australia, spot fires have been
reached to the 25 km-away from main line front. The spotting fuels consist of
small cones, piece of bark, grass clumps and other such fuels. Once spotting
begins, it is very difficult to take under control. Spot fires extensively increase the
rate of fire spread and because of this, many fire fighters have been trapped in
unexpected situations of fire and they have been injured, even lost their lives
(Edmonds et al., 2000).
2.3. The Impacts of Forest Fires
The main injurious to forest protection and ecosystem is forest fires. The
first visible and immediate impact of forest fires are destruction to human
communities and forest ecosystems. On the other hand, the potential adverse
effects of the forest fires are supply of ecosystem services necessary for the health
of human communities. Disorder of forest ecosystem causes increase in the soil
erosion, decrease in the infiltration capacity of the soils, loss of valuable soil
nutrition and change in the quality and quantity of water, both in streams and
16
ground water. Defeat of biological diversity due to forest fire alters the species
composition and distribution even causes species extinction.
Some of forest fires injuries can be classified under the following heads:
injury to trees containing merchantable material and young growth, injury to soil,
injury to the productive power of the forest, injury to recreational and scenic
values, injury to wildlife, injury to forage, injury to human life and others.
During fire, trees take damage at the scale of trivial wound at the base to
complete consumption of the tree. Tree death occurs when the cambium or living
layer between bark and wood of the tree is killed. If the trees that have an
economic value are killed due to fire, they should be cut and utilized before the
timber decays or is attacked by insects. It is strongly advised that merchantable
trees injured but not killed by fire should be removed because decay and insects of
trees frequently increase the final loss of the timber consumed by the fire. Many
years must elapse in order to become as a merchantable tree. During this time
growing trees can be infected by insects and fungi from the fire scars and finally
within one or two years, the trees practically become worthless after fire.
Forest fires affect the physical properties, particularly penetrability and
porosity of soil more critically than its chemical properties. “Physical properties
of soils are influenced by fires through decrease in the humus content.” When the
humus in the soil is once influenced, the repeated fires seriously injure the humus
content of the soil. The forest litter contains nutritive materials mainly, nitrogen,
17
calcium, phosphorus and potash. The most important material is nitrogen because
it is volatile and lost to the soil when fire consumes the forest litter.
In some places forest areas grows on the rocky areas only a few inches
below the surface. This thin soil layer that consists of largely organic materials
completely consumed due to fire. As a consequence, the soil valuable for forest is
entirely destroyed.
Another fire injury to the forest is the productive power of the forest. The
composition of the forest type is changed due to fire. Many of the valuable tree
species is more sensitive to fire than others and fire kills the sensitive species and
leaves the resistant ones and the more resistant and less valuable species form the
dominant vegetation cover of the area. If frequently happens, brush and woody
shrubs cover the burned area. Sometimes fire destroys whole vegetation cover and
becomes barren or the trees must be cut because of fire scared. Therefore the
capability of producing good timber crops has been decreased.
The most visible injury of the fire is demolishment of the recreational and
scenic values of the area. Community income from the tourism is adversely
affected because of fire. Also, fires affect the wildlife directly and indirectly.
Many animals can not escape and actually burned to death. An indirect effect of
the fire is destruction of the food and cover that is important for animals as
shelter. Fire destroys plants and dry grass that have a forage value. When the
18
intensity of fire is sufficient, fires kill the roots of the plants, reducing the density
of stocking.
The most valuable duty of the forests is protection of the areas against
erosion, avalanches, landslides and shifting sands. As well as fire adversely
affects the nature and ecosystem, it also threat the property of human beings,
buildings, live stocks, farms and even towns, and even their lives.
Beside of adverse effects, forest fires have also beneficial effects if the
burning is occurred under control. Fire can be used as a tool for establishing
natural reproduction of the desired species. Fire can improve the physical
condition of the soil by aerating and increasing its temperature. Additionally, fire
may be useful for controlling plant diseases. The most important beneficial use of
prescribed fire is prevention of more destructive fires (Hawley and Stickel, 1948).
2.4. The Origins of Forest Fires in Turkey
Fire is a natural and usual part of the ecosystem; however it is strongly
affected by the human activities both directly and indirectly. The reasons of the
forest fires change according to the society, activity that is performed at the wild
area. The variableness of the reasons of the forest fires influences the time of the
fires.
19
In accordance with 10-year data, 47 % of the forest fires are caused by
negligence, carelessness and accident, 13 % are incendiary, 6 % is caused by
lightning and 34 % is due to unknown reasons. In 2002, the 55 % of the forest
fires were started because of negligence, carelessness and accident, 15 % of them
are incendiary, and 12 % is due to lightning (Figure 2.5). However, the reason of
18 % of the 2002-forest-fire is unknown (General Directorate of Forests in
Turkey, 2002).
CARELESSNESS 55%
UNKNOWN 18%
INCENDIARY 15%
LIGHTNING 12%
Figure 2.5. The Reasons of the Forest Fires in Turkey in 2002
Among the wildfires whose source is negligence, carelessness and
accident, 9.7 % of the fires and 6 % of the burned areas are due to debris burning,
8.5 % of the fires and 45.6 % of the burned area are Camp Fire burned by
shepherd and lastly 9 % of the fires and 1.8 % of the burned area in 2002 are
owing to smokers.
In the last few years, the number of wildfires caused by energy transfer
lines increased. 3 % in 1997, 15 % in 1998, 2 % in 1999, 5.1 % in 2000, 2.5 % in
2001 and 3.3 % of the fires were burned in 2002 and in addition to this,
20
respectively, 3 %, 15 %, 7.5 %, 19.7 %, 5.3 % and 24.8 % of the total burned area
have been lost annually from 1997 to 2002 (General Directorate of Forest in
Turkey, 2002).
According to Hawley and Stickel (1948), the causes of the fires is
classified as camp fire, debris burning, that is fires originally set for clearing land
for any purpose, incendiary, lightning, lumbering, railroad, smoker and
miscellaneous. The classification method of fire reasons is used by most fire
control organizations and grouping fire reasons in the same way for all the regions
in the country and it is an advantage for fire management.
In conclusion, human beings are the main reason of forest fires directly or
indirectly. 90 % of the total fires are man-caused and could be prevented.
2.5. Forest Fires in Turkey
27 % of Turkey’s lands are covered by forest (Figure 2.6). 48 % of this
forest areas are productive however, 52 % must be protected. The number of total
forest fires between 1937 and 2002 is 72316 so the number of yearly fires has
been found as1096.
Between 1993 and 2002, 21000 forest fires have been occurred and the
yearly average has been calculated as 2100 (Figure 2.7). In 2002, the first 5
regions where the most of the fires have occurred can be listed as Muğla (232
21
fires), İzmir (232 fires), Amasya (194 fires), Istanbul (170 fires), Antalya (166
fires) and Balıkesir (153 fires).
Figure 2.6. The Overall View of the Forest Areas in Turkey (The color, Green
shows forest areas, brown shows DEM) (http://www.ogm.gov.tr)
2545
3239
17701645
1339
19322075
2353
2631
1471
0
500
1000
1500
2000
2500
3000
3500
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
Time (Years)
Num
ber
of F
ires
Figure 2.7. The Distribution of Number of Forest Fires between 1993-2002
(General Directorate of Forest in Turkey, 2002)
Mean Value Line
22
From 1937 to 2002, 1,549,506 hectare area have been burned and
according to this, around, 23,477 ha area has been destroyed annually (Figure
2.8). If the data of last 10 years have been examined in accordance with the
Region Head Offices, Muğla (1958 ha), İzmir (1475 ha) and Antalya (1352 ha)
have taken the first 3 places in ranking. In 2002, the most damaged area belonged
to Balıkesir (3634 ha), Muğla (2072 ha), Antalya (450 ha), Mersin (421 ha) and
İzmir (309 ha).
15393
38128
7676
14922
6316 6764 5804
26353
7394 8514
0
5000
10000
15000
20000
25000
30000
35000
40000
45000
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
Time (Years)
Bur
ned
Are
a (h
a)
Figure 2.8. The Distribution of Burned Area between 1993-2002 (General
Directorate of Forest in Turkey, 2002)
The forest fires have been classified and defined according to the burned
area by General Directorate of Forest in Turkey. The classes of the forest fires are
denoted as A, B, C, D, E, F, G1, G2 and G3 in accordance with the size of burned
areas in an ascending order (Table 2.3).
Mean Value Line
23
G3-class forest fires, have occurred in Balıkesir-Kepsut, where 3573 ha
forest area has been lost, and Marmaris – Çetibeli, where 1775 ha forest area has
been lost in 2002. 62.8 % of the total area has been destroyed in these forest fires.
C, D, E and F class medium forest fires formed the 26 % of the total area.
Although the percentage of A and B class forest fires was 92 according to the
number of fires, the burned area formed the 11.1 % of the total area. The number
and area of the forest fires between 2000 and 2002 according to the sizes can be
seen in Table 2.3.
Table 2.3. The Classes of Fires According to Sizes (General Directorate of
Forest in Turkey, 2002)
Number of Fires Area of Fires Classes
2000 2001 2002 2000 2001 2002
A (<1 ha) 1602 1907 1108 548 636 343
B (1.1 – 5.0 ha) 507 531 246 1361 1451 607
C (5.1 – 20.0 ha) 136 143 90 1465 1471 635
D (20.1 – 50.0 ha) 55 31 8 1898 1044 279
E (50.1 – 200.0 ha) 34 15 11 3293 1146 1020
F (200.1 – 500.0 ha) 9 2 1 3236 515 280
G1 (500.0 – 800.0 ha) 2 2 1184 1131
G2 (800.1 – 1500.0 ha) 4 4471
G3 (>1500.0 ha) 4 2 8897 5348
TOTAL 2353 2631 1471 26353 7394 8514
General Directorate of Forest that consists of 27 head offices manages the
forests in Turkey. These offices are divided according to the density of the forest
areas (Figure 2.9). These forest areas were degreed from 1 to 5 in accordance with
24
Figure 2.9. The Head Offices of General Dircetorate of Forest in Turkey
25
the fire risk. Degree 1 shows the most sensitive areas and decrease to Degree 5.
Figure 2.10 and Figure 2.11 show the fire risk consistent with all type of forest
areas and red pine forest areas. Therefore, the distribution of the forest fires
depends on the fire risk and can be seen in Figure 2.12 and Figure 2.13 according
to number and area of fires in 2002.
2.6. Using Geographic Information Systems in Fire Management
Forest fires are spatial phenomena so they can be modeled using geo-
spatial technologies, such as geographic information systems (GIS) and remotely
sensed high resolution imagery. Buckley (2001) provides a review of applications
in remote sensing and GIS technologies for fire and fuel management. In this
work, vegetation of United States Marine Corps. Camp Lejeune are mapped to
Northern Forest Fire Laboratory (NFFL) fuel models. The elements of wildland
risk and hazard are defined, modeled and mapped levels of concern allowing for
regular updating due to changes in fuels and land use. Jaiswal et al. (2002) used
GIS for combining different forest fire causing factors for determining the forest
fire risk zone map. “Forest fire risk zones were delineated by assigning subjective
weights to the classes of all the layers according to their sensitivity to fire or their
fire-inducing capability” (Jaiswal et al., 2002). It was concluded that GIS-based
forest fire risk model was in strong agreement with actual fire affected sites.
Hardwick (1999) evaluated the GIS the use of GIS technology as a tool in
reducing large fire costs, and to identify the benefits of GIS to incident
management efforts. Both quantitative and qualitative benefits were examined and
26
Figure 2.10. The Distribution of Forest Areas According to Fire Risk Degrees in 2002 (Data Source: General Directorate of Forest in
Turkey, 2002)
27
Figure 2.11. The Distribution of Red Pine Forest Areas According to Fire Risk Degrees in 2002 (Data Source: General Directorate of
Forest in Turkey, 2002)
28
Figure 2.12. The Distribution of Forest Fires According to Number of Fires in 2002 (Data Source: General Directorate of Forest in
Turkey, 2002)
29
Figure 2.13. The Distribution of Forest Fires According to Area Burned in 2002 (Data Source: General Directorate of Forest in
Turkey, 2002)
30
it was found that GIS would be a useful tool for large fire management.
Fire prediction systems model the fire behavior using site specific data,
which are weather, topography, fuel type and condition. Albright and Meisner
(1999) listed four types of fire prediction models: physical, physical-statistical,
statistical and probabilistic.
Physical models are based on the physics of heat transfer and combustion,
for example Albini (1986) (Albright and Meisner, 1999). It is not currently used
due to the requirement of huge amount of data. The most commonly used models
are physical – statistical fire prediction models. They combine the physical theory
with statistical correlation. Rothermel’s (1972) model, approximating the solution
from laboratory experiment, and The Canadian Forest Fire Behavior Prediction
(FBP) System (Forestry Canada Fire Danger Group, 1992 in Albright and
Meisner, 1999) are examples of physical – statistical models. Statistical fire
predication models use equations derived from test fires. Their success depends
on the conditions similar to the test fires. McArthur’s fire danger meters
(McArthur, 1966 in Albright and Meisner, 1999; Noble et al., 1980) are the
examples of statistical fire prediction models. The last fire prediction models are
probabilistic that are based on contingency tables (Albright and Meisner, 1999).
Some of the GIS – based fire simulation systems can be counted as
FARSITE (Finney, 1998), FIREMAP (Ball and Guertin, 1992 in Albright and
Meisner, 1999), WILDFIRE (Wallace, 1993 in Finney, 1998) and DYNAFIRE
31
(Kalabokidis et al., 1991 in Albright and Meisner, 1999). These systems are based
on physical- statistical model. Except for FARSITE, all of simulation systems
have been developed for simulating the low to moderate intensity surface fires.
FARSITE can simulate the spread and behavior of wildland for both surface and
crown fire. The outputs of all prediction systems include GIS vector and raster
files with different format (Albright and Meisner, 1999).
Hawkes et al., (1997) have developed a prototype wildfire threat rating
system (WTRS) via Geographic Information System (GIS) technology. This
system is based on fire risk, values at risk, fire behavior, and suppression
capability. They have been determined as a function of 13 factors, which are
Digital Elevation Model, aspect, road, etc. Wildfire threat consists of four main
components and a number of factors contribute to each of these components.
A fire simulation application, called as FIRE!, has been developed by the
Weinstein et al., (1995). It integrates the fire behavior modeling into the
ARC/INFO GIS environment. “FARSITE interacts seamlessly within the
ARCINFO environment as a component of FIRE!...” (Weinstein et al, 1995). At
this study, forest fire behavior model integrates the spatial 13 type NFFL Fuel
Model, topographic data, temporal weather and wind settings and initial fuel
moistures into the prediction of forest fire behavior across both time and space.
Liu and Chou (1997) provide a cell automation method to simulate wildfire
growth by using a grid – based GIS. Rothermel’s model was used to estimate the
rate of spread and fire intensity.
32
Vega et al., (1998) have developed specific equations to predict rate of
spread and flame length for Galician shrublands. “Fire behavior data recorded
from field experimental fires carried out in Galician shrublands have been used to
compare predictions from different existing models to observed data.” Also
Fernandes (2001) has developed a preliminary model to predict rate of spread in
shrub fuel model by analyzing the relationship between fire spread and
environmental variables in Portugal.
FARSITE has been used by Noonan (2002) in order to test the
effectiveness of fuel treatments in a shrub and timber area in the Eastern Sierra
Nevada. It has been concluded that the treated fuel decreases the dangerous fire
behavior, especially shrub. Fujioka (2001) simulated the Bee Fire with given
landscape, fuel and weather conditions. The aim of the simulation is describing a
new methodology to evaluate the uncertainties of fire spread simulation. The
resultant fire spread simulations were moderately successful comparing with Bee
Fire. In addition to these studies, Van Wagtendonk (1996) have tested the various
fuel treatments in mixer conifer vegetation via FARSITE model. Fuel treatment
scenarios have been formed by changing the fuel model values for load and depth,
in other words defining custom fuel models. Also Stephens (Stephens, 1995 in
Van Wagtendonk, 1996) used for testing fuel treatments for protecting the
Tuolumne Grove of giant sequoias at the head of North Crane Creek in Yosemite
National Park. Removals of ladder fuels and salvage logging with and without
slash treatment have been tested as moderate intensity of burn. His study increases
33
the importance of fuel treatments for example prescribed burning and defensible
fuel profile zones in critical areas.
34
CHAPTER 3
FARSITE: FIRE AREA SIMULATOR MODEL
Forest fires which occur in the arid and semi-arid regions of the world
have disastrous social, ecological and economic impacts, causing loss of life and
property, loss of vegetation, oil and water resources. Computers and electronic
devices are increasingly being employed in spotting, monitoring and combating
forest fires, providing assistance to the tools traditionally used in fire
management.
In this chapter, two-dimensional computer simulation model, FARSITE,
that incorporates existing fire behavior models for surface, crown, spotting, point
source fire acceleration is discussed. Application areas, capabilities, and data
requirements are briefly explained. After that, limitations and assumptions are
listed.
Fire Area Simulator, FARSITE, is a deterministic computer based program
designed to model the spread and behavior of fires spatially and temporally under
conditions of heterogeneous terrain, fuels and weather. It was based on the
implementation of Huygens’ Principle of wave propagation for simulating the
35
growth of a fire front. Fire behaviors and perimeters are manageable numerically,
graphically and spatially to other PC and GIS applications. However, in order to
get these advantages, the more organized and GIS based spatial data on the
topography, fuels and weather data are required. The PC version of FARSITE
requires the GIS software ARC/INFO or GRASS to generate, manage and provide
spatial data layers, these are fuels and topography (Figure 3.1) (Finney, 2003;
Finney and Andrews, 1999). Finney and Andrews (1999) state that “FARSITE is
widely used by the State and Federal agencies as well as private parties in the
United States, who recognize the value of having GIS-based data on fuels and
vegetation for a variety of applications.”
“The Microsoft Windows interface offers flexibility for office or field
prediction of fire growth. Fire growth and behavior scenarios can be developed
relatively quickly using short term weather forecasts or long term weather
projections (ideally based on historic records)” (Finney, 2003).
3.1. The Application Areas
The FARSITE model was initially developed to support the management
prescribed natural fires (pnfs), which is now called as fire use and it was aimed for
operational and planning parts. The National Wildfire Coordinating Group
(NWCG), which coordinates interagency federal fire management, defined the
wildfire as “…an unwanted wildland fire” and prescribed fire as “…the
purposeful application of fire under predetermined conditions of fuels, weather,
36
and topography to achieve management objectives” (Edmonds et al., 2000). The
terminology prescribed natural fires (pnfs) was defined as “…naturally ignited
ignition that is allowed to burn under an approved management plan that includes
monitoring” (Edmonds et al., 2000). In the new terminology, term of prescribed
natural fires is redefined as a managed wildland fire.
Figure 3.1. Raster Input Layers Generated by the Geographic Information
Systems for FARSITE Simulation (Finney, 1998)
According to Edmonds et al., (2000) it is used to support decision making
for park and wilderness managements. “Based on expected or historic weather
patterns, fire spread and intensity can be simulated over time, providing
information for monitoring strategies, fire suppression strategies, assigning
priorities to multiple fires, and other uses. It can be used to simulate either
prescribed fires or wildfires, and has important forest health management
implications for planning” (Edmonds et al., 2000). It is stated in Finney (2003)
37
that the most relevant intended usage are short and long term simulation of active
and potential fire.
The FARSITE model has three main uses, which are simulation of past
fires, active fires and potential fires. Simulation of past fires assists to understand
how well it generates known fire growth model and fire spread pattern according
to given available data. “Simulating past fires is critical in developing confidence
for using FARSITE to project the growth of active fires.” (Finney and Andrews,
1999)
FARSITE was originally developed for long – range active prescribed fires
on national parks and wilderness areas. 1 or 2 – day short – range projections on
large fires have also been simulated and simulation results can be used to support
strategic firefighting decisions. Moreover, partial section of fire fronts can be
simulated for immediate interest (Finney and Andrews, 1999). According to
Finney (2003), the intended usage of active fires have been described as preparing
daily fire situation analysis for short term projections; forecasting fire monitoring
activities based on fire line arrival, fire behavior and fire effects; defining long
term fire variability and range of outcomes and lastly assigning the priorities to
multiple fires for both short and long term projections.
The most appropriate and common application of FARSITE is fire
planning activities. For example, they include analyzing fire management
alternatives and examining suppression opportunities for fires that start in various
38
locations or weather scenarios. Potential fires at different locations can be
examined under a variety of fuel and weather conditions (Finney and Andrews,
1999). Actual weather records following suppression could be used for simulating
the likely burn extent and behavior for planning later prescribed burning of the
area at a more suitable time” (Finney, 2003).
During fire suppressions, comparison of the effectiveness of different
strategies under varying weather scenarios and suppression efforts with and
without fuel treatment programs can be performed by FARSITE. Lastly, cost of
suppression strategies can be estimated (Finney, 2003).
3.2. Capabilities
FARSITE uses raster data as inputs for all spatial data on topography,
fuels, and crown structure. It can also export fire behavior information as raster
themes to use in a GIS.
Van Wagtendonk (1996) has stated that utilizing the spatial database
capabilities of Geographic Information Systems, FARSITE allows the users to
simulate the spatial and temporal spread and behavior of the fire over
heterogeneous terrain, fuels and weather. Therefore it is more realistic and
accurate modeling of actual fire growth, as well as the efficiency and capacity for
investigating effectiveness of fuel treatments design to diminish hazard. In
39
addition to this spotting and crowning model allows to provide an ideal tool for
investigating extreme fire behavior.
Finney (1998) states that FARSITE calculates the surface fire behavior,
crown fire behavior, fire acceleration, and fuel moisture. The surface fire model
developed by Rothermel (1972) is connected to the Van Wagner (1977; 1993) and
Rothermel (1991) crown fire model in order to simulate the movement of fire to
crown. Torching tree model developed by Albini (1979) is used to simulate
spotting distance (Finney and Andrews, 1999).
FARSITE produces the fire growth perimeters and fire behavior maps in
ARC/INFO, ArcView, and GRASS GIS based formats. Fire perimeter outputs are
in vector format as either line or polygon. Raster maps are also produced to show
fire behavior at each cell in any resolution. They include Time of Arrival, Fireline
Intensity, Flame Length, Rate of Spread, Heat per Area, Reaction Intensity,
Crown Fire Activity and Spread Direction. While GIS based output maps are
produced, also table and graph based output data are generated.
Environmental and combustion raster maps can also be generated by
FARSITE. Environmental maps display fuel moistures and weather conditions
across the landscape at a specific time. It is useful to understand the effects of
topography, forest canopy structure, and weather. In addition to this,
instantaneous fire activity can be recorded and viewed for all areas within the fire
area by using combustion map (Finney, 2003).
40
FARSITE can simulate the fire suppression by means of ground and aerial
attack tactics. Ground attack tactics consist of direct, indirect and parallel attack.
Direct attack is applied as following the instant perimeter of the fire using data on
fireline production rate according to crew type and fuel model. Indirect attack is
building impermeable fireline along a predetermined direction. Parallel attack is
similar to direct attack except for building fireline at a specific constant distance
from the fire front. Air attack features allow the user to locate the retardant drops
for a defined aircraft (Finney and Andrews, 1999).
3.3. Data Requirements
FARSITE requires several types of data files within GIS environment.
Required data consist of three legs of the fire triangle (Figure 2.1). Fuel and
topography are required as spatial GIS based raster themes whereas weather data
are formatted as stream or ASCII formatted values. Eight cell-based layers consist
of elevation, slope, aspect, canopy cover, fuel model (Anderson, 1982), live crown
base height, crown bulk density, and tree height or canopy height. The first five
raster data sets are required in order to run FARSITE (Finney, 1998).
FARSITE requires the support of Geographic Information System to
generate, manage and provide spatial data themes as GRASS ASCII or ARC
GRID ASCII format and they are combined into a single landscape (*.LCP) file in
FARSITE. All themes must be co-registered in order to have the same reference
point, projection and units. Identical resolution of raster themes must be the same
41
and finally all raster data sets must be in the same extent (Finney, 2003). Spatial
themes are provided in raster format because of providing rapid access by the
model to the necessary spatial data. 25 to 50 m raster resolutions are fairly enough
to provide accurate level of detail (Finney, 1998).
Optional data themes are required only to calculate some aspects of fire
behavior, such as crown fire and fuel consumption. Surface fire simulation can be
performed without optional GIS data themes (Finney, 2003)
FARSITE also requires weather and wind data in order to perform
simulations. FARSITE input data files are tabulated in Table 3.1.
3.3.1. Landscape File
According to the Finney (2003), the extension of the Landscape File is
.LCP and it contains all the rasterized data themes imported from GIS. It includes
5 basic themes; elevation, aspect, slope, fuel model, canopy cover and optional
files for stand height, crown base height, crown bulk density, duff loading, and
course woody profiles. These optional parameters can be assigned constant unless
they are provided.
A landscape file is a binary file and it consists of a header and a body of
short integers for each theme. The header contains information about the bounds
of area, cell resolutions and units of each theme.
42
Table 3.1. FARSITE Input Data Files (Finney, 2003)
File Name File Ext.
File Type Required Optional
Landscape .LCP Raster Fuel Model, Slope, Aspect, Elevation, Canopy Cover
Crown Bulk Density, Crown Base Height, Stand Height, Duff Loading and Coarse Woody themes
Weather .WTR Text At least one file Use up to 5 .WTR files in a simulation
Wind .WND Text At least one file Use up to 5 .WND files in a simulation
Adjustment .ADJ Text Although required, this file can consist of all zeros
Adjustment factors other than zero are optional
Initial Fuel Moisture
.FMS Text Needs moistures at least one day before the beginning of the simulation
none
Fuel Model Conversion
.CNV Text none yes
Custom Fuel Models
.FMD Text none For fuel models other than the 13 standard NFFL models
Fire Acceleration
.ACL Binary none yes
Air Attack Resources
.AIR Text none Needed to implement the air attack functions
Coarse Woody Profiles
.CWD Text none Specifies > 3" fuels for the Coarse Woody GIS theme used by Post Frontal Combustion Model.
Burn Periods .BPD Text none Specifies a daily burn period by date
Gridded Weather and Winds
.ATM Text none Uses gridded weather files if a weather model to provide them is available
Ground Attack Resources
.CRW Text none needed to implement the air attack functions
43
The landscape file is generated by FARSITE through obtaining each
ASCII grid files and specifying their units.
i. Fuel Model Theme
Fuels consist of the various components of live and dead vegetation that
occur on a site. Type and quantity of fuels depend on the soil, climate, geographic
features, and the fire history of the site.
The mathematical model of fire behavior requires the description of fuel
properties as inputs to calculation of fire behavior. The collection of fuel
properties is known as fuel models and it is divided into four groups: grass, shrub,
timber, and slash. “The differences in fire behavior among these groups are
basically related to the fuel load and its distribution among the fuel particle size
classes” (Anderson, 1982). Fuel models are the tools that help the user to estimate
fire behavior realistically.
Fuel model theme must consist of integer index to a fuel model. Between
Model numbers 1 and 13 were reserved for standard NFFL Fuel Model. Detailed
information on NFFL Fuel Model is given in Appendix A (Anderson, 1982).
Numbers 0, 98 and 99 are used for non fuel raster cells, such as water and rocky
ground.
44
Any fuel model other than standard 13 NFFL fuel models (Anderson,
1982) can be described in a Custom Fuel Model (.FMD) file, which is described
in Section 3.3.2.
ii. Slope Theme
One of the required raster themes is the slope theme and its values should
be integer although decimal values can also be read. Slope units can be degrees or
percent of inclination from horizontal and it is needed in order to figure slope
effects on fire spread and solar radiance (Finney, 2003).
iii. Aspect Theme
The aspect theme is one of the required raster themes and contains values
for topographic aspect. If the aspect data are from a GRASS ASCII file all values
are oriented counterclockwise from east. Aspect values from ARC/INFO must be
azimuth values (degrees clockwise from north) and can be integer or decimal
values (Finney, 2003).
iv. Elevation Theme
Elevation theme can have units of either meters (m) or feet (ft) above sea
level. In order to adjust adiabatic temperature and humidity, this theme is required
and it is necessary for conversion of fire spread between horizontal and slope
45
distances (Finney, 2003). In addition, slope and aspect themes are derived form
elevation theme.
v. Canopy Cover Theme
One of the essential themes is named as canopy cover due to compute
shading and wind reduction factors for all fuel models. Canopy cover is the
horizontal percentage of the ground surface that is covered by tree crowns and it is
measured as the horizontal fraction of the ground that is covered directly overhead
by tree canopy (Finney, 2003).
The units of canopy cover theme can be selected either categories (1 to 4)
or percentage values (0 to 100). Percentage values can be categorized as
1. 1 – 20 %
2. 21 – 50 %
3. 51 – 80 %
4. 81 – 100 %
and zero cover is assigned by 0 or 99.
Furthermore, as well as fuel model, canopy cover can be assigned as a
constant value across the whole landscape. Since canopy cover theme controls
how FARSITE uses other themes, it is important and required in order to run
46
simulation. If the value of a cell in the canopy cover theme is assigned zero,
FARSITE understand that there is no tree cover at this location and then it ignores
the values of stand height, crown base height, and crown bulk density for the same
cell (Finney, 2003).
vi. Stand Height Theme
Stand height theme is one of the optional spatial data themes and it is used
to compute spotting distances, wind reduction to midflame height, and crown fire
characteristics. Stand height is displayed in Figure 3.2. The values can be either
integer or decimal in the units of meters or feet. The precision of fire behavior
calculations is limited to 1/10th of the units (Finney, 2003).
Figure 3.2. Definition of Crown Fuel Data Used in FARSITE (Finney, 2003)
47
vii. Crown Base Height (CBH)
Crown base height (CHB) is important for determining transition from
surface fire to crown fire. It is the height to the bottom of the live crown (Figure
3.2). CHB should include ladder fuels effects in increasing vertical continuity and
assisting shift to crown fire. The precision of crown base height is same as stand
height theme (Finney, 2003).
viii. Crown Bulk Density (CBD)
Crown bulk density (Figure 3.2) is used to determine the spread
characteristics of crown fires. The values of raster themes can be integers or
decimal values in either kg/m3 or lb/ft3. “The precision of crown bulk density
values are used to the nearest 100th kg/m3 or 1000 lb/ft3. Thus, if you have integer
values in your crown bulk density theme, you must use the calculator features of
your GIS to multiply the original crown bulk density units by 100 (kg/m3) or 1000
(lb/ft3). This conversion is optional for decimal bulk density values” (Finney,
2003).
There are two options for generating crown bulk density information. The
first one is that Crown Bulk Density is linked to the canopy cover theme. In this
method, only maximum crown bulk density for each forest type is entered and
canopy cover is assumed to be 100 %. During the simulation, actual crown bulk
density for each raster cell is computed by the multiplication of entered crown
48
bulk density and canopy cover fraction. The second option is using independent
crown bulk density theme developed by GIS (Finney, 2003).
Moreover, there are also two optional raster themes, namely Duff Loading
and Coarse Woody, which are necessary for utilization of the Post-Frontal
Combustion model. “Post-Frontal combustion refers to the burning of woody
surface fuels, litter, and duff behind the moving forward edge of the flaming
zone” (Finney, 2003).
Raster file data units and raster themes in order to generate landscape file
and their usage were summarized in Table 3.2 and Table 3.3.
Table 3.2. Raster File Data Units (Finney, 2003)
File Theme Required Default Units Alternate Units
Elevation yes Meters feet
Slope yes Degrees percent
Aspect yes categories 1-25 degrees
Fuel model yes 13 NFFL (Anderson,
1982) models
custom or converted
models
Canopy cover yes categories 1-4 percent
Tree height no meters*10 meters, feet, feet*10
Crown base height no meters*10 meters, feet, feet*10
Crown bulk
density
no kg/m3*100 kg/m3, lbs/ft3,
lbs/ft3*100
Duff loading no Mg/ha Tons/acre
Coarse woody no coarse woody
models
none
49
Table 3.3. Raster inputs to FARSITE and their usage in the simulation (Finney,
1998)
Raster Theme Units Usage Elevation m, ft Used adiabatic adjustment of temperature and
humidity from the reference elevation input with the weather stream.
Slope percent Used for computing direct effects on fire spread, along with Aspect, for determining the angle of incident solar radiation (along with latitude, date, and time of day) and transforming spread rates and directions from the surface to horizontal coordinates.
Aspect ° Az See slope
Fuel Model Provides the physical description of the surface fuel complex that is used to determine surface fire behavior (see Anderson 1982). Included here are loadings (weight) by size class and dead or live categories, ratios of surface area to volume, and bulk depth.
Canopy Cover percent Used to determine an average shading of the surface fuels (Rothermel et al., 1986) that affects fuel moisture calculations. It also helps determine the wind reduction factor that decreases windspeed from the reference velocity of the input stream (6.1 m above the vegetation) to a level that affects the surface fire (Albini and Baughman, 1979).
Crown Height m, ft Affects the relative positioning of logarithmic wind profile that is extended above the terrain. Along with canopy cover, the influences of the wind reduction factor (Albini and Baughman, 1979), the starting position of embers lofted by torching trees, the trajectory of embers descending through the wind profile (Albini, 1979).
Crown Base Height
m, ft Used along with the surface fire intensity and foliar moisture content to determine the threshold for transition to crown fire (Alexander, 1988; Van Wagner, 1977).
Crown Bulk Density
kg m-3, lb ft-3
Used to determine the threshold for achieving active crown fire (Van Wagner, 1977; 1993).
50
3.3.2. ASCII Input Files
ASCII text file consists of weather (.WTR), wind (.WND) and Fuel Files;
these are Adjustment (.ADJ), Moisture (.FMS), Conversion (.CNV), Custom
Models (.FMD) and Coarse Woody (.CWD). Weather, wind, adjustment, and fuel
moisture text files are required. In this section, the information about these input
files is presented.
The format of all ASCII input files are space delimited and it can be
created and edited via text editing application such as Microsoft Notepad or by
using FARSITE editor. The parameters and its definitions of ASCII input files
and also examples have been given in Appendix B.
i. Weather (.WTR)
The weather file is a required ASCII text file in order to perform any
simulation. A weather file must contain daily observations on temperature,
humidity and precipitation specified below. All input must be integer value in
English or metric units.
The usage of weather information is to interpolate temperature and
humidity for hours between the maximum and minimum values of each day. Also
these data are extrapolated to different elevations by using the elevation theme in
the landscape.
51
Five weather streams can be used and this feature permits to approximate
some spatial variation in weather. In order to use multiple weather stations, some
overlaps in each weather file must exist (Finney, 2003).
ii. Wind (.WND)
As well as weather file, wind file is also one of the required input files and
can be as a stream of data in a wind file (.WND) or as a gridded weather (.ATM)
file, described in gridded weather and winds section. And also up to 5 wind files
can be used as input for a project to simulate spatially varying winds, for example
ridge winds vs. slope winds.
Although winds vary in space and time, FARSITE assumes that winds are
constant in space for a given wind stream but variable in time. In other words,
there are no topographic effects on winds. The wind inputs can be at any temporal
resolution, hourly or sub-hourly and it must be integer. FARSITE allows weather
inputs in English or metric units, stated at the first line of the Wind (.WND) file.
Additionally, wind observation can be on irregular interval, for example every 10
minutes during day and only every 2 hours at night (Finney, 2003).
iii. Adjustment Factors (.ADJ)
Adjustment Factors File is required in order to adjust spread rate of fire
according to the user experienced judgment or local data to tune the simulation to
52
observed or actual fire spread patterns (see Section 3.4). “Factors are fuel model
specific and are multiplied by the rate of spread to achieve the specified
adjustment” (Finney, 2003).
Only rate of spread for a simulation is modified by adjustment factors. In
order to change the all fire behavior parameters, heat per unit area, fire intensity
etc., Custom Fuel Model must be created by using adjustment factors.
Additionally, adjustment factors for each custom model used must be specified
(Finney, 2003).
iv. Initial Fuel Moisture (.FMS)
“The fuel moistures at the beginning of the simulation must be set for each
fuel type. These fuel moistures are required to begin the process of calculating site
specific fuel moistures at each time step throughout the simulation.” If custom
models are used, they too must have initial fuel moistures specified in this file
(Finney, 2003).
v. Fuel Model Conversion (.CNV)
Fuel conversion file is an optional ASCII text file. If the fuel model
numbers in the raster fuel model theme do not match directly to the 13 NFFL
(Anderson, 1982) or custom fuel models files, Fuel Model Conversion (.CNV)
53
File can be used. Another usage is for changing the fuel model that a particular
fuel attribute represents.
vi. Custom Fuel Models (.FMD)
Custom Fuel Model is an optional ASCII text file for FARSITE. Fuel
models different than standard 13 NFFL fuel model (Anderson, 1982) must be
defined in a Custom Fuel Model (.FMD) File and they are assigned model
numbers 14 through 50. The unit of fuel model input can be either English or
metric, that is inserted at the first line of custom Fuel Model (.FMD) File. Its
parameters are specified below (Finney, 2003).
Individual custom fuel models can be tuned by changing the fuel model
parameters. When using adjustment file, only the spread rate are modified. In
order to adjust fire behavior parameters, fuel models must be tuned by using
tuning feature of FARSITE (Finney, 2003).
vii. Fire Acceleration (.ACL)
Fire Acceleration (.ACL) File allows controlling point and line fires
acceleration for each fuel type. “Fire acceleration is defined as the rate of increase
in spread rate/fire line intensity from a given source.” The type of acceleration
definition files are in binary format so that they can be modified only by using the
Fire Acceleration command in FARSITE (Finney, 2003).
54
viii. Air Attack Resources (.AIR)
Information of aerial fire fighting resources is stored in Air Resource
(.AIR) File. These files include the capabilities of different aircraft (air tankers,
helicopters dropping water or retardant). More than one aircraft information can
be defined in one file. For each aircraft, specific name have to be assigned. The
resource name must begin and end with character ‘#’.
Line length is the altitude of retardant drops for each coverage levels.
Coverage levels need to be higher in heavier fuels to be effective. For line length,
to acquire realistic numbers in fuel types, experience and/or consultancy of
literatures are needed (Finney, 2003).
ix. Coarse Woody Profiles (.CWD)
Coarse Woody Profile is an optional ASCII text file required to run the
Post-Frontal Combustion Model in FARSITE. The file has profile numbers
between 1 and 99. The units are defined by the words ENGLISH or METRIC at
the first line of file. The file can contain multiple lines per profile.
The header consists of two lines. The first line starts with the word
“MODEL”, the fuel model number (integer) and brief description of the model.
The second line starts with the word “DEPTH” and the depth of the profile
(decimal) in feet or centimeters (Finney, 2003).
55
x. Burn Period (.BPD)
The burn period file specifies daily burning periods. It is an optional file
that consists of dates and times of burns. It is used to stop the simulation during
periods of low activity to correct the inclination to over predict fire growth rates
when environmental conditions essentially stop fire spread.
This feature also speed up simulations by stopping the simulation during
periods of low fire behavior that affects the result of fire behavior slightly
(Finney, 2003).
xi. Gridded Weather File (.ATM)
FARSITE accepts gridded inputs for weather and wind variables. Weather
variables are temperature, humidity and precipitation; wind variables are wind
speed and direction, and also cloud cover.
Data units are same as weather and wind stream inputs for FARSITE to
allow compatibility with both GRASS and ARC GRID data. The six gridded input
files are temperature (Fahrenheit), humidity (Percentage), Precipitation (Inches
×100), Wind Speed (MPH), Wind Direction (Azimuth Degrees), and Cloud Cover
(Percentage). Formats of all gridded input files are similar to traditional GRASS
or ARC GRID ASCII raster. The filenames of the gridded input files must be
included into an Atmosphere (.ATM) File to define the date and time of each
56
raster files. The format of the atmosphere file changes according to using gridded
inputs for either weather and wind variables or only wind variables (Finney,
2003).
Examples of Gridded Weather (.ATM) File can be seen in Appendix B.
xii. Ground Resources File (.CRW)
Ground resources file (.CRW) is an optional ASCII file needed to simulate
fire suppression with ground resources. Ground base fire fighting resource
information is defined in this file. These files describe the capabilities of different
resources, these are had crews, engines, equipments etc.
3.4. Limitations and Assumptions
Some assumptions had to be made to model fire growth. Models are valid
and useful if the scopes of the assumptions are clearly understood. The following
paragraphs explain the major assumptions of the models used in FARSITE.
1. “The shapes of fires are elliptical under uniform conditions.” (Finney,
1998) This assumption was verified by Van Wagner (1969). Different
fire shapes have been reported by Peet (Peet, 1967 in Finney, 1998),
Albini (1976), Anderson (Anderson, 1983 in Finney, 1998), Alexander
(Alexander, 1985 in Finney, 1998) and Byram (Byram, 1959 in
57
Finney, 1998). Richards (Richards, 1993 in Finney, 1998) analyzed
that none of fire shape alternatives could explain variation in
windspeeds or direction of spread. Richards (Richards, 1993 in Finney,
1998) assumed that “fire spread was independent of the shape or length
of the fire front”. An elliptical fire shape assumption is true only in
continuous fuels so fire shapes due to discontinuous fuels will not be
effectively modeled.
2. “Fire spread rate and intensity at a given vertex is assumed to be
independent of fire and environmental interactions.” (Finney, 1998)
according to Cheney et al., (Cheney et al, 1993 in Finney, 1998) and
Weber (Weber, 1989 in Finney, 1998), field observations and analysis
suggest that flame length of line fire affects the spread rate and fire
shape.
3. “Fire acceleration is assumed to be dependent on fuel type but
independent of fire behavior.” (Finney, 1998). This means that time
period must be constant for given fuel type in order to reach steady-
state spread rate regardless of environmental conditions. Acceleration
rate for each fuel type must be assigned (McAlpine and Wakimoto,
1991 in Finney, 1998).
58
4. “Fires are assumed to instantly achieve the expected elliptical shape
when burning conditions change (such as changes in windspeed or
slope)” (Finney, 1998).
5. “The elliptical shapes are assumed to be fuel independent, meaning the
fire shape (not size) is only determined by the resultant wind – slope
vector” (Finney, 1998).
6. “The origin of an elliptical fire is assumed to be located at the rear
focus of the ellipse.” (Finney, 1998) Most of the model using elliptical
fire (Alexander, 1985 in Finney, 1998; Anderson, 1983 in Finney,
1998; Andrews, 1986) accept this assumptions due to providing an
implicit means to determine backing fire spread rate.
7. “The spread of a continuous fire front can be approximated using a
finite number of points.” (Finney, 1998). The accuracy of this
assumption depends on the spatial resolution determined by the user
and the resolution specified for the simulation. “It is assumed that a
resolution can be specified that preserves the "important" features of
fire growth but ignores irrelevant spatial detail. This is dependent on
the purpose and requirements for the simulation” (Finney, 2003).
8. FARSITE model has not been developed to determine whether a fire
will spread or not. It has not been only designed for spread of fire but
59
also spread rate results can be adjusted by the use of adjustment and
custom fuel models (Finney, 2003).
60
CHAPTER 4
THE CASE STUDY: MARMARİS- ÇETİBELİ WILDFIRE
Fire Area Simulator Model, FARSITE, is used in modeling the fire
behavior in order to develop fire management strategies. The suitability of the
model for modeling the wildfires in Turkey is tested and the appropriate fire
model parameters are obtained.
In this chapter, the forest fire, which started in Marmaris Çetibeli on the
15th of August and lasted till 27th of August, 2002, has been simulated. The
information about the simulation sites and wildfire, the preparation of the required
input data of FARSITE have been explained. Digital elevation model (DEM) of
the area has been generated from 1:25,000 scale maps by digitizing contours. The
weather and wind data have been gathered and put into appropriate GIS layers.
Afterwards, the 13 Northern Forest Fire Laboratory Fuel Model of the area have
been determined according to vegetation map, which was prepared by General
Directorate of Forest. The IKONOS images acquired on the dates April 22, 2002
and August 21, 2002 are used to define the fuel model of the area by using visual
interpretation. The production of FARSITE input data files is explained briefly.
61
Since the exact fuel model is not available for Turkey, the model is used to
calibrate the fuel model belonging to the study area. The burned area is extracted
form the fire reports and IKONOS images.
4.1. The Study Area
The Aegean and Mediterranean Regions, covered by red pine (Pinus
brutia) forests, are the most sensitive areas to fire in Turkey. The red pine forests
located around Çetibeli District in Marmaris have burned 3 times, these are the
largest forest fires occurred in this area, as presented in Table 4.1. Since it is a fire
occurred recently, the documentation of the fire is better than the others.
Therefore, the Çetibeli fire started at 15th of August, 2002 is studied in this study.
The study area is located in Muğla, in the south border of Aegean Region,
the position of burned area is about 1 km south east of Çetibeli district and it is
elongated with west to east direction. The main road, named D 400, passes from
west of the region. Elevations vary from 10 to 600 m along North West to South
East direction. The location of the study area can be seen in Figure 4.1.
The climate in Çetibeli is warm like Mediterranean type, which is cold
rainy winters, short wet springs and autumns, and hot dry summers. However,
general climatic characteristics vary locally in the high mountainous areas. The
area is mainly covered with red pine (Pinus brutia), which is the dominant
vegetation species of the Mediterranean Region.
62
Table 4.1. The Largest Forest Fires in Turkey (General Directorate of Forest in
Turkey, 2002)
Directorate of
Region
Directorate of
Management
Management Chief
Start Date
End Date
Burned Area (ha)
Reasons
Adana Osmaniye Osmaniye 9/30/99 10/2/99 1200 Incendiary
Balıkesir Bandırma Aladağ 4/5/00 4/6/00 1267 ETL1
Muğla Marmaris Hisarönü 8/11/97 8/13/97 1385 Carelessness
Muğla Aydın Söke 7/27/96 7/30/96 1438 Unknown
Denizli Denizli Buldan 7/13/00 7/14/00 1459 Carelessness
Çanakkale Keşan Çınarlıdere 9/1/00 9/3/00 1689 Carelessness
Antalya Antalya Düzlerçam 7/21/97 7/22/97 1715 Carelessness
Muğla Marmaris Çetibeli 8/15/02 8/17/02 1776 ETL
Bursa Orhaneli Merkez 4/5/00 4/6/00 1970 Carelessness
Antalya Taşağıl Karabük 8/3/00 8/5/00 2102 Incendiary
Adana Karaisalı Akarca 8/3/00 8/6/00 3138 Carelessness
Balıkesir Balıkesir Kepsut 8/12/02 8/13/02 3573 Unknown
Çanakkale Çanakkale Eceabat 7/25/94 7/27/94 4049 Debris burning
Muğla Marmaris Çetibeli 7/27/96 8/1/96 7090 Carelessness
Muğla Marmaris Çetibeli 3/23/79 10/2/79 13260 Unknown
i. Information about Çetibeli Wildfire
The information about Çetibeli forest fire was reported on 11st of
November, 2002 by the Directorate of Marmaris Forest Management (Marmaris
Orman İşletme Müdürlüğü) and Çetibeli Forest Management Chief (Çetibeli
Orman İşletme Şefliği). Starting time of fire is 15th of August, 2002 at 18:23. It
started at Çetibeli Forest Management Region and passed to the Gökova Forest
Management Region on 16th of August, 2002 at 16:45. The fire is reported by the
people who live in Çetibeli District and Altınsivri Watch Tower (Figure 4.1).
1 Energy Transmission Line
63
Figure 4.1. Location of the Study Area
64
The first interference was made at 18:30 on 15th of August, 2002 by fire
fighters. The starting location of the fire was near transformer, that is about 90 m
distant to the asphalt road and the slope of the starting location is about 20 %. Fire
started as surface fire and continued as surface and crown fire due to strong wind.
The rate of spread of fire was about 900 m /hr and intensity of fire was very high.
The flame height was measured about 20 to 2500 cm. The vegetation types of the
forest were red pine trees and dense bushes. This area has been burned before in
1955, 1979 and 1996 (Table 4.1). The data representing the meteorological
conditions during fire are tabulated in Table 4.2.
The equipments used to suppress fire consist of scratch, some sort of
knifes, saws, dozers, helicopter and planes. The fire fighters have tried to suppress
the fire by means of direct, indirect and parallel attack. The squads of Marmaris,
Muğla, İzmir, Antalya, Denizli, Mersin, Bursa and Ankara Forest Management
have joined the fire battle. In addition to this, some planes that have come from
İstanbul, Çanakkale and İzmir and two military planes have participated into fire
suppression by dropping retardant water.
Table 4.2 Meteorological Conditions during Fire (obtained from Muğla
Meteorological Station)
Maximum Minimum
Temperature (°C) 34 24
Humidity (%) 70 35
Wind Direction and Speed (km/h) 5-6 N NW 2-4 N NW
65
The number of people who have attended in order to suppress fire was
arranged as 3 managers, 62 forest engineers, 40 park rangers, 1125 fire fighters,
300 volunteers, 300 soldiers, 15 police officers and 30 gendarmes. The used
vehicles and machines are 17 planes, 13 helicopters, 27 dozers, 25 trailers, 1
excavator, 1 break down lorry, and 83 fire trucks. The registry report of fire is
given in Appendix C.
Finally fire was taken under control on 17th of August, 2002 at 21:00 and it
was suppressed completely on 27th of August, 2002 at 18:30. It took 288 hours
and 7 minutes. The total burned area was measured as 1775.5 hectares.
On the other hand, the starting location of the fire is not known precisely.
The progressing time and location of the fire, the local weather and wind data, and
the information about suppression tactics, types, time, number of workers and
squad and the interfere location of ground and aerial forces during fire fighting
have not been reported. Moreover, the feature which could be valued as fire
barrier should have been noted.
ii. Fire Cost Analysis
The cost analysis has been made by the Çetibeli and Gökova Forestry
Chief on 13rd of September, 2002. The total cost was tabulated in Table 4.3.
66
Table 4.3. Total Cost of Çetibeli Fire (Marmaris – Çetibeli Forest Management
Chief, 2002a)
Costs (×106 TL)Tree Cost 233,154 TLReforestration Cost 3,107,125 TLFeeding Cost 22,985 TLWorker Cost 213,990 TLTimber Cost 233,519 TLGasoline Cost 33,712 TLDozer Cost 42,450 TLTrailer Cost 4,100 TLSaw Cost 500 TLPlane Rent Cost 331,335, TLHelicopter Rent Cost 1,054,397 TLFire Truck Cost 126,000 TLTotal 5,403,267 TL $ 3,235,4892
4.2. The GIS Data Preparation
i. The Preparation of GIS Data Files for Çetibeli
In this section, the preparation steps of required data for Çetibeli are
explained. The optional landscape input files of FARSITE could not be generated
due to lack of information and impossibility of measurement at field. Registration
and generation of GIS themes have been performed by using ArcGIS 8.1 and its
extensions, which are Spatial Analyst and 3D Analyst. First of all, topographic
maps of the area were registered to UTM ED 1950 zone 35N coordinate system.
The 3rd order polynomial transformation method was used and they were rectified
by using Bilinear Interpolation. Each topographic paper map was georeferenced 2 1$ = 1,670,000 TL
67
using a set of 30 ground control points (Figure 4.2). The detailed coordinate
system parameters, the link table of each registered maps and all of layers created
for simulation can be seen in Appendix D. The contours, roads, rivers and energy
transfer lines were digitized from the rectified topographic maps.
ii. Selection and Processing of Satellite Images
NFFL Fuel Types (Anderson, 1982) of the study area have been classified
according to the vegetation map of the study area and pre-fire satellite image. The
vegetation map has been drawn by the General Directorate of Forest. IKONOS
image acquired on 22nd of April, 2002 was used as the pre-fire satellite image. The
radiometric resolution of the post and pre-fire images are 11-bit, the visible and
near infrared bands have 4 m spatial resolution and the panchromatic bands have 1
m spatial resolution. The image acquired on 21st of August, 2002 is used as post-
fire image. The overall view of the images can be seen in Figure 4.3 and Figure
4.4. The satellite images do not cover all of the study area (Figure 4.1).
ERDAS Imagine 8.5 software was used in order to handle satellite images.
For both images, red, green, blue and near infrared layers of each image were
stacked into single file and spatially enhanced by merging with 1-m resolution
panchromatic image. The projections of the images are UTM and their datum and
spheroid are World Geodetic System 1984 (WGS 84). The datum and spheroid of
the images are transformed to the European Datum 1950 and International 1909.
68
Figure 4.2. Ground Control Points on the Topographic Maps Belonging to the
Study Area
69
Figure 4.3. Pre-fire IKONOS Satellite Image, (22nd of April, 2002 at 09:05) (A)
Visible Spectrum, (B) Near Infrared
70
Figure 4.4. Post-fire IKONOS Satellite Image, (21st of August, 2002 at 09:15) (A)
Visible Spectrum, (B) Near Infrared
71
iii. Identification of Fuel Types and Canopy Cover
Northern Forest Fire Laboratory (NFFL) in United States of America has
classified the vegetation types into 13 standard fuel models. FARSITE uses 13 -
NFFL fuel types as fuel model themes (Anderson, 1982).
In Turkey, forest areas were divided into sections according to terrain or
geographic features such as rivers. These sections were separated into subsections
whose vegetation types show the same characteristics, cover density and age and
they are named as special codes such as Çzc1 etc. These codes are tabulated in
Appendix E. The dominant vegetation species of the study area was red pine
(Pinus brutia) with different cover density and age. Some part of the area covered
with bushes and field, that is, open land. Generally, identification of the fuel types
was built upon vegetation maps and satellite images. The vegetation map of
Çetibeli has been obtained from General Directorate of Forest. This vegetation
paper map has been scanned into digital format and registered via collecting
reference points from topographic maps. After registration process, each polygon
in vegetation map has been digitized and the vegetation codes have been stored in
its database. Entire fuel map of the study area has not been digitized because
digitized area is sufficient for performing simulation and it covers the burned area.
According to NFFL fuel model description (Appendix A) the vegetation of
the study area was tried to be estimated by the help of the pictures of the fuel
models, vegetation map and experiences of fire engineers and it was checked via
72
satellite image visually. Since the satellite image does not cover entire study area,
only the covered part has been verified. Fuel types of uncovered parts have been
assigned by the help of estimated areas and vegetation map. Sorting the records by
common name helped to speed up the categorization process (Finney, 2003).
“Canopy cover is measured as the horizontal fraction of the ground that is
covered directly overhead by tree canopy” (Finney, 2003). Coverage units can be
categorized with numbers 1 – 4 or percentage values of the cover. The main
source of the classification of the canopy cover was vegetation map and satellite
images. Canopy cover of the subsections was decided as 1 – 4 categorization scale
due to codes, which were assigned by the forest engineers who have drawn the
map and they were compared with satellite image.
4.3. Generation of Landscape Input File
The landscape file consists of 5 themes; elevation, slope, aspect, canopy
cover and fuel model. In order to generate elevation theme, Digital Elevation
Model (DEM) of the terrain was generated by using spatial interpolation of
digitized contour lines from the 1: 25,000 scale (contours every 10 m) paper maps.
Kriging method was performed in order to interpolate to raster. However, DEM
could not be generated from polyline contour coverage hence Triangulated
Irregular Network (TIN) of the area was produced and TIN was converted to point
features by using 3D Analyst. After that, the coverages, elevations of top points
digitized from paper maps and elevation generated from TIN, were merged by
73
Geoprocessing Wizard and the final elevation coverage was produced. Ordinary
Kriging method was selected and the Search Radius Type was fixed. The default
Output Cell Size was calculated 23.14533128 meter but it is fixed to 20 meters.
According to ArcGIS Desktop Help, the Search Radius Distance should be 5
times greater than output cell size. The parameters of the Kriging interpolation
method were reported in Lineage Report in Appendix D (Figure 4.5).
Slope and aspect of the area were derived from DEM of the area by means
of Spatial Analyst Extension of ArcGIS. The output cell size was the same as
DEM and output unit of slope was selected as degree. Aspect of the area was
calculated as azimuth angle from the North. Elevation, slope and aspect of the
work area can be seen in Figure 4.5.
After fuel models and canopy cover of the area have been decided, all of
these vector features were converted to raster by using fuel model and canopy
cover fields separately with the same cell size, 20 m. Overall view of fuel models
and canopy cover are presented in Figure 4.6. Finally all of these landscape
themes, elevation, slope, aspect, fuel model and canopy cover were exported to
ASCII Raster format via ArcView GIS 3.2 with Spatial Analyst extensions.
However, entire work area (Figure 4.1) have not been converted into ASCII
format. FARSITE can be run if all of the required data themes of the area exist.
Hence, before exported to ASCII Raster format, elevation, slope and aspect
themes of the area have been clipped according to the spatial extent of fuel model
theme. It has been dissolved by Geoprocessing Wizard into one polygon and
74
Figure 4.5. Data Themes of Landscape File. (A)Elevation, (B) Slope, and (C)
Aspect of Çetibeli
75
Figure 4.6. The (A) Fuel Model and (B) Canopy Cover of the Study Area
converted to raster format. By using raster calculator, clipping has been carried
out by multiplying both themes. The important point is that all data themes of
landscape file must have the same cell size, and spatial extent.
The vector themes can also be used as fire barrier and improvement of
visibility.
76
4.4. Creating ASCII Input Files
FARSITE uses some required and optional input files. All of these input
files are ASCII text format so they can be created, edited or viewed by any text
editor such as Microsoft Notepad. The required ASCII input files are weather,
wind, adjustment, and initial fuel moisture.
Meteorological data was obtained from Turkish State Meteorological
Organization (Appendix F). The name of station is Marmaris located at 28.16°
latitude, 36.51° longitude (Figure 4.1) and it has an elevation of 16 m from sea.
The data taken from Meteorological Organization can be listed as air temperature
(hourly), wind direction (16 directions) measured hourly, and wind speed (m/s),
humidity (%) and cloudiness (%) recorded 3 times (7:00, 14:00, 21:00) in a day.
Wind data are measured at 10 m high from the ground.
Weather data consist of precipitation, minimum and maximum
temperature and their record time, minimum and maximum humidity, elevation
and precipitation time period with daily observation. The starting date of the
weather file must be one day before the fire starts because it is needed to calculate
initial fuel moisture of the fuel. The weather data were generated by using
meteorological data.
77
Weather and wind data were prepared by using software Microsoft Excel
in order to facilitate formatting and converting to appropriate unit. Afterwards,
tabulated data were exported to ASCII text files.
Adjustment Factors File was left with default values at first because it is
used for adjusting the rate of spread of fire. The final adjustment factors are
achieved after the calibration is performed by running the simulation until the
expected result is acquired.
The ASCII input files of Çetibeli are documented in Appendix G
4.5. Simulation Process of Çetibeli Wildfire
The information of Çetibeli Wildfire has been summarized in Section 4.1.
In this section, the Çetibeli Wildfire has been simulated and according to the
known information of the fire, real situation has been tried to be set in order to
obtain the real fire spread and behavior.
By the help of Fire Evaluation Reports (2002b) and satellite images, the
information about burned areas, ignition points and the spread location at a
specific time have been extracted. The ignition point has been reported as being
near the transformer building. By the help of these information and satellite
image, the location of the ignition point has been estimated as North – West of
burned area near Çetibeli Village, illustrated in Figure 4.7.(A).
78
The burned area has been plotted on vegetation map by forest engineers,
who joined the fire suppression. By the help of post-fire satellite image, the
burned area covered by satellite image has been digitized. However, the east part
of the burned area has been digitized according to the sketch which shows the
burning sections. The burned area and the ignition point are illustrated in Figure
4.7.(B).
Figure 4.7. The View of (A) Ignition Point and (B) Burned Area
79
The time and location of the fire was known while passing from Çetibeli
Forest Management Region to Gökova Forest Management Region on August
16th, 2002 at 16:45 and the fire has been simulated until this time at that location.
Since the passing way was not known, it was assumed that fire spread has reached
at this location as surface and crown fire.
The FARSITE simulation consists of a series of preparation steps. First
step is preparation of spatial and nonspatial data as input. Other is selecting model
settings and options and the last one is controlling the simulation process (Figure
4.8).
FARSITE requires and generates many types of data files. In order to
prevent confusion, two directories, “input” and “output”, were created. Input
directory contains GIS themes, ASCII input files, project, landscape and
bookmark files. Output Directory contains the product of simulation.
i. Building Landscape (.LCP) File
The Landscape (.LCP) File is generated by inputting the ASCII raster files.
Five required themes have been loaded and correct units have been selected. In
this project, SI units are used. The latitude of the work area has been entered as 37
N. Once landscape file was created, there were no need to load ASCII raster
themes. Figure 4.9 shows the landscape file generation window of Çetibeli
Project.
80
Figure 4.8. Flow Diagram of FARSITE (Finney, 2003) Model
81
Figure 4.9. Landscape (.LCP) File Generation.
ii. Generation of FARSITE Project (.FPJ) File
A Project (.FPJ) File contains the spatial, i.e. landscape file, and nonspatial
inputs of the simulation. The project file has been saved at the same directory of
input files as required. It has been generated because it speeds up the loading of
input files and their settings. However, it does not contain any run time
information about starting and ending time, ignition location, parameters and
options. Figure 4.10 illustrates the Project Inputs; Landscape, Adjustment, Fuel
Moisture, Weather and Wind Files. All ASCII files have been given in Appendix
G.
82
Figure 4.10. FARSITE Project File Generation
After generating project file, simulation can be initiated. In order to start
simulation, the parameters and options of the simulation are set and the location of
ignition points and the output files are selected.
iii. Model Parameters Settings
The spatial and temporal detail of the fire behavior calculations are
determined by the Model Parameters (Finney, 2003). There are no correct settings
because the suitability of the model settings depends on the purpose of the
simulation. Model parameters must be set for each simulation. Model parameters
83
are time step, visible step, perimeter resolution and distance resolution (Figure
4.11).
Figure 4.11. Model Parameters
The definition of the time step is the maximum amount of time that the
conditions at a given point are assumed constant in order to calculate position of
fire front. The positions of all fires are calculated at this time step so that possible
merges can be computed (Finney, 2003).
Time step is inversely proportional to the fire spread rate. Finney (2003)
states the time step values of surface fire as approximately 30 to 120 minutes for
timber fuels, 10 to 20 minutes for brush and dry grass. For extreme surface fires or
torching / crowning fires, time step may be set as 5 to 10 minutes for all fuel
types. According to these proposals, time step has been set as 30 min.
84
The time step has the secondary importance when compared with the
spatial resolution of the calculation. “The internal time step used by the simulation
is constantly changing according to the minimum time required for the fire to
spread the distance equaling the distance resolution. The actual time step is thus,
only used as a consistent period during which all fires will be projecting to a
coincident time before mergers and spotting are computed” (Finney, 2003).
Visible time step is only used for the graphical representation of fire
spread and behavior. It is always a multiple of the actual time step in order to
avoid unnecessary temporal detail in fire perimeter positions. Setting of primary
visible time step is required but secondary visible time step is optional and only
used to differentiate fire growth at two meaningful time periods. The fire front is
drawn according to these intervals and visible time step affects the time resolution
of vector output file from the simulation. The primary and secondary visible time
steps have been set as 30 minutes and 1 hour, respectively.
Perimeter resolution determines the maximum distance between points
used to define the fire perimeter. It is defined as resolution of a fire front in the
direction tangential to the fire perimeter at each point. Setting of perimeter
resolution is essential due to the determination of the amount of landscape
information used in the simulation. In addition to this, it controls the detail of the
fire front to respond to heterogeneities occurring at a fine scale.
85
Meaningful perimeter resolution can be set by means of practical
knowledge gained through mapping of actual fire fronts. Finney (2003) states that
logical perimeter resolution are not more than about twice the raster resolution of
30 m. The raster resolution of Çetibeli inputs are 20 m so perimeter resolution has
been set as 30 m.
The distance resolution means the maximum spread distance from any
perimeter point. “This distance can not be exceeded in a time step before new
fuels, weather and topography data are used to compute the spread rate.” The
distance resolution can not be greater than the perimeter resolution due to the
detection and elimination methods of crossovers. FARSITE calculates and
automatically adjusts the distance resolution. The value that has been set in Model
Parameters is the maximum value of distance resolution. “The most logical value
for distance resolution would be approximately the same as that for the
perimeter.” (Finney, 2003) Therefore, distance resolution has been entered 30 m.,
same as perimeter resolution.
All of the model parameters are illustrated in Figure 4.11.
iv. Duration Settings
Duration of fire must be set for every FARSITE simulation after weather
and wind file have been loaded. The starting time of the Çetibeli fire simulation
has been set on 15th August at 18:23 and ending time has been set on 16th August
86
on 16:45 (Figure 4.12). Fuel conditioning period is an option to adjust fuel
moistures before the starting of the simulation. “When the Use Conditioning
Period for Fuel Moistures check box is selected, FARSITE starts with values in the
Initial Fuel Moisture (.FMS) File and calculates fuel moistures across the
landscape based on elevation, aspect, slope, and shading before the Starting Time
of the simulation” (Finney, 2003). If conditioning periods are not used, fuel
moistures at every point across the landscape are used from the Initial Fuel
Moisture File. The effect of conditioning period is different in shorter simulations
with longer timelag fuels. Conditioning period has been checked at this
simulation.
Figure 4.12. Simulation Duration
v. Fire Behavior Options Settings
Fire Behavior Option Settings controls the burn methods. Fire behavior
calculations and other option, especially crown and spot fire settings, can be
87
modified. Simulation of crown fire can be enabled or disabled depending on the
crown fuel conditions and surface fire behavior. In order to simulate crown fire
accurately, the Crown Base Height (CBH), Crown Bulk Density (CBD) and Stand
Height Themes should exist in the Landscape File. However, constants for crown
fuels can be defined for entire landscape to simulate crown fire. Selecting this
option does not force to occur crown fire. Only it allows the FARSITE model to
determine if transition occurs. Initially, only the surface fire has been simulated at
the first run so this option has been remained unchecked.
Crown Bulk Density is critical parameter to determine crown fire spread.
If constant Crown Bulk Density value is used, in order to modify Crown Bulk
Density at each cell, link crown density and cover option are checked. In Figure
4.13, the settings of fire behavior options have been illustrated.
Figure 4.13. Fire Behavior Options
88
vi. Ignition Point Settings
There are two primary methods to locate the ignition source of simulation.
First one is locating by the ignition mode on map. Next one is importing a point,
line or polygon vector file. The ignition location of Çetibeli fire has been
discussed before. The ignition location Arc Shape file has been imported.
Usually there are barriers to prevent fire spread on terrain such as
highways, roads, or streams. Barriers can be located like ignition, manually or
importing vector file. Any barrier has not been set initially.
vii. Output Settings
The other information except for area and perimeter calculation, perimeter
can be exported as vector and fire behavior parameters are exported as raster files.
Raster files can be viewed in FARSITE or GIS and remote sensing applications.
There are four functions of output settings.
• Setting units of outputs as Metric or English,
• Determining the file names and formats of the vector fire perimeters,
• Selecting the eight different fire behavior characteristics raster files, their
name and resolution
89
• Option of creating output log files with information on the parameters
used, simulation inputs, and start/end times for each vector or raster file
created (Finney, 2003).
Table 4.4. Raster Output Themes Filename Extensions and Units (Finney,
2003)
Parameters Extensions Metric English
Time of Arrival .TOA hours hours
Fire Line Intensity .FLI kW/m BTU/ft/sec
Flamelength .FML m ft
Rate of Spread .ROS m/min ft/min
Heat per Unit Area .HPA kJ/m2 BTU/ft2
Reaction Intensity .RCI kW/m2 BTU/ft2/sec
Crown-NoCrown .CFR 1=surface,
2=passive,
3=active
1=surface,
2=passive,
3=active
Spread Direction .SDR 0-359 ° Azimuth 0-359 ° Azimuth
All units have been selected as Metric. Fire perimeters have been exported
as polygon Arc Shape file. Only Crown Fire Activity raster output file has not
been selected because crown fire was not simulated. Raster fires have been
exported as GRID ASCII format at 20 meter resolution. The export and output
options can be seen in Figure 4.14.
Except output files, the tabulated output data and graphs can be also
exported to text files. If simulation is suspended at any time, the visible fire
perimeter being generated by the current simulation can be saved.
90
Figure 4.14. Export and Output Options of Çetibeli Fire Simulation
After all of these steps have been completed, Çetibeli Fire was ready to be
simulated.
4.5.1. Çetibeli Fire Simulation Run
In this section, Çetibeli fire has been simulated by changing simulation
parameters in order to approach the real fire spread. All of the simulation
parameters log file generated by FARSITE can be seen in Appendix H. The fire
attack inputs and parameters could not be set because of the real situation,
location and information about crews, equipments and vehicles are unknown. The
91
view of the FARSITE before and during simulation run has been illustrated in
Figure 4.15 and Figure 4.16.
Figure 4.15. The View of FARSITE Before the Run
After every simulation has been completed, visible fire perimeters of
vector files and fire behavior raster files, which are Time of Arrival, Fireline
Intensity, Flame Length, Rate of Spread, Heat/Area, Reaction Intensity and
Spread Directions, have been generated. Also area and perimeter table of fire at
each elapsed time have been saved. All of raster and vector GIS output projections
have been defined by ArcToolbox 8.1. ArcGIS 8.1. is used for analysis and
demonstration of outputs.
92
Figure 4.16. The View of FARSITE During the Run
Only the surface fire has been simulated at the first run. According to the
real fire information, the fire should have spread up to the end of the work area
during fire duration. However, the fire could not reach the expected point at the
end of the simulation. The simulated burned area has been calculated as 339 ha
and only 115 ha of the real burned area could be estimated correctly; displayed as
hatched in Figure 4.17. 224 ha area has been over estimated. The ratio of correctly
estimated area to the actual burned area is obtained as an accuracy measure of the
simulation. In this run, the accuracy is calculated as 9%. Under these
circumstances, the simulation of the fire should be improved.
93
Figure 4.17. Simulated and Real Burned Area (First Run)
The raster outputs of fire behavior have been also generated at the end of
the run. In addition to this, some optional raster outputs, fuel moisture, air
temperature, mid flame wind, relative humidity and solar radiation of the terrain
have been produced. These output files can be generated at any time during
simulation. The fire behavior themes can be seen in Figure 4.18. In this figure, air
temperature theme shows the temperature of the air on August, 16th at 16:45.
4.5.2. Calibration and Adjustment of Çetibeli Fire Simulation
Interpreting the output of the simulation and comparing the real fire
information suggest that simulation of Çetibeli fire growth needs to be improved.
Finney (2003) defined the calibration as “the process of diagnosing problems and
making improvements to the simulation, usually compared to observations of
actual fire behavior”.
94
Figure 4.18. Fire Behavior Outputs. (A)Time of Arrival, (B) Fireline Intensity, (C)
Flame Length, (D) Rate of Spread (E) Heat Per Unit Area, (F) Spread Direction,
(G) Air Temperature
95
Calibration is necessary for many reasons. These reasons can be
categorized into 3 caption; input data, parameters and settings, fuel and fire
behavior models. Selected models may not reflect the reality of fire or the suitable
model may not be selected. The variability and inaccuracies in input parameters
and settings are reasons to do calibration (Finney, 2003).
At the second run, crown fire model is also simulated with surface fire.
According to the fire registry reports, crown fire exists and it must be simulated
with surface fire although optional raster GIS themes required for crown fire
simulation, Crown Base Height, Crown Bulk Density and Stand Height, are not
available. Hence constant values have been used. Average height of red pine trees
was assumed as 10 m, so Stand Height and Crown Base Height values were
entered as 10 and 6 m. Crown Bulk Density value was selected as 0.2 kg/m3.
FARSITE generated these optional themes according to their constant values.
In Fire Behavior Options, crown fire option was enabled and “Link Crown
Density & Cover” option was checked. By this option, Crown Bulk Density was
generated in respect of canopy cover because Crown Bulk Density was set as
constant but some areas on the landscape was not covered by trees.
After simulation has been performed, the total simulated burned area has
been calculated as 424 ha. 135 ha of the burned area (Figure 4.19) have been
estimated correctly. Accuracy of the model run is calculated as 10 %. However,
96
overestimation of burned area is 289 ha. Fire spread rate is still slow according to
the real spread rate and almost similar to the first run spread rate.
Figure 4.19. Simulated and Real Burned Area (Second Run)
Rate and direction of fire spread depend on mainly the fuel model and
wind speed. Wind speed and direction can not be edited because it has been
measured by the station. The only parameter that can be changed was fuel model.
The adjustment factors adjust the spread of each fuel model but they do not affect
the fire behavior.
At the third run, the simulation has been performed after adjustment
factors and acceleration value of fire have been changed. When fuel model map
has been examined, the real burned area covers mainly the fuel model 1, 3 and 9.
Hence adjustment factors of these fuel models have been changed as 0.3, 3.8 and
3.8, respectively. Acceleration values have been adjusted in accordance with the
97
previous runs because fire had to reach the east boundary of the work area at the
end of the time.
The total simulated burned area has been determined as 1409 ha. 70 % of
the real burned area (911 ha) has been estimated correctly but approximately 500
ha of simulated burned area has been overestimated. Additionally, at the end of
the simulation, firefront has almost reached to the east boundary of work area,
illustrated in Figure 4.20.
Figure 4.20. Simulated and Real Burned Area (Final Run)
In order to generate the fire behavior outputs, the custom fuel models have
been defined according to the adjustment factors because they only affect the
spread rate. Custom fuel models have been created by modifying the basic fuel
models by using adjustment factors. Rather than regenerating fuel model as a
raster theme, Conversion Factor (.CNV) File has been used. The modified ASCII
input files have been illustrated in Appendix G.
98
After modification has been completed, the fire simulation has been
performed once more. The fire spread and fire behavior outputs have been
generated. The total simulated burned area has been calculated as 1210 ha. 70 %,
of the real burned area has been estimated correctly. 360 ha of the area have been
simulated incorrectly. The fire spread has been illustrated in Figure 4.21.
Figure 4.21. Simulated and Real Burned Area (Final Run after Modification)
In addition to this, time of arrival, fire line intensity, rate of spread and
spread direction of the fire can be seen in Figure 4.22 and Figure 4.23.
4.6. Discussion of the Results
After the calibration of Çetibeli fire has been completed, the outputs of the
fire behavior, time of arrival, rate of spread, spread direction, and time contour of
fire spread have been obtained. Fire behavior outputs could not contain the real
99
Figure 4.22. The Outputs of (A) Time of Arrival, (B) Rate of Spread and (C)
Spread Direction.
100
Figure 4.23. The Outputs of (D) Relative Humidity, (E) Air Temperature and (F)
Solar Radiation
101
fire information because there is not any documented information about fire
behavior of Çetibeli Fire. The only criteria for comparison are spread rate and
direction and also shape of burned area.
According to the fire spread (Figure 4.21), some parts of the burned area
are overestimated and some parts are underestimated. These problems lie with
inaccurate data of fuel moistures, fuel models of the area and weather data. Wind
reduction factors of forested areas and extreme topographic variations, such as
sheltering effect on the spread rate of fire in some parts of the landscape. This
result indicates how sensitive the fuel model and moisture parameters in the fire
simulation modeling.
Fire has different spread directions according to the real burned area. In
order to suppress the fire, some interference and attacks have been done by the
fire fighters and machines. It was previously mentioned that the location, time of
attacks to the fire and information about number of crews and vehicles were not
known, so attack simulation could not be performed. Suppression activities affect
the fire spread rate, shape and behavior. Because these parameters could not enter
for the simulation as input, overestimation in estimating the burned area has
occurred. Also fuel model of the area could not be defined properly.
The reasons of underestimation could be due to fuel type and
meteorological conditions those do not reflect the real conditions. In addition to
this, the types of fire, surface and crown, could not be simulated appropriately
102
because in order to simulate crown fire, some optional landscape data themes such
as crown bulk density, crown height are needed. During simulation, these themes
were entered as constant values. Also this could be the reason for overestimation.
Finney (1998) states that the disparity of real fire and simulated fire is due
to the variance of the scale between the frequency of data inputs to the simulation
and the frequency of variation in real environmental conditions affecting a fire. “If
the scale of input data to the simulations is much coarser than that of the real
environment (fuels, weather and topography), the fire behavior equations will tend
to produce equilibrium values rather than reflect the cumulative outcome of
fluctuating fire behavior” (Finney, 1988).
When the time contour of fire spread (Figure 4.24) and Figure 4.22.(B)
have been examined the fire spread was slow initially. This could be due to the
fuel types and meteorological data. Meteorological, i.e. weather data should have
been measured on site during fire. Fuel Model 15, modified from Fuel Model 3,
burned more rapidly than the others. It is obtained that Fuel Model 15 may be the
dominant vegetation of the study area.
The spread of fire is almost similar to the real fire (Figure 4.24). The
compatibility of the fire model is acceptable because in spite of lack of
information and input data about fire, the accuracy of the simulation in estimating
the burned area after calibration has been found as 70 %.
103
Figure 4.24. Time Contour of Fire Spread (30 min interval) and Fuel Models
The most important input of fire model is fuel model theme. Fire behavior
highly depends on fuel model. The second important input is weather data. It
should be measured on site and recorded with sufficient time period. The last
important input is topography. Approximately 30 m resolution DEM is sufficient
for fire modeling. Some topographic features affect fire behavior unexpectedly.
This problem can be overcome by experience.
104
CHAPTER 5
CONCLUSIONS AND RECOMMENDATIONS
In this thesis, the suitability of a GIS – based forest fire simulation model
has been tested and the requirements of the model have been determined for
Turkey by simulating a past fire which occurred in Marmaris - Çetibeli.
Analysis of past fires has been performed for checking the performance
and accuracy of the model by means of comparing the satellite images and output
maps of the model. During active fires, ground and aerial suppression tactics can
be planned more accurately and more rapidly depending on the results of fire
simulation models. In addition to this, future planning could be performed easily
for forest areas which are extremely sensitive to fire.
The study area has been decided according to previously occurred
wildfires and suitable vegetation, weather and topographic data have been
obtained. The IKONOS images of the study area have been obtained and burned
areas were extracted from them.
105
The necessary input files of fire model have been generated. Firstly,
contour lines of the study area have been digitized. Elevation, slope and aspect
layers of study area have been created. Vegetation layer of the study area has been
produced from vegetation maps and fuel models of the area have been decided by
using vegetation maps and IKONOS Image. Weather data have been obtained
from meteorological station and prepared into suitable format.
After preparation of input file, the simulation of both surface and crown
fire has been run and the result of the model showing fire spread area was
compared with the extracted burned area. Fire did not reach the expected location
at the given time duration. This result showed that calibration was needed. Fuel
model has been adjusted according to the simulation results. Finally, 70 %
accuracy of the fire model run has been achieved and new fuel model parameters
have been defined for the study area. In addition to this, fire behavior outputs such
as fire intensity, direction, rate of spread have been obtained.
In this study, it is found out that the fuel model and fuel moisture
parameters are the most sensitive parameters of the model. The dominant
vegetation type indicated as BÇzY in forestry maps was represented as fuel model
15.
Retrieving the vegetation information from satellite images having high
spatial resolution increased the accuracy of the vegetation maps. Hence the
determination of fuel models from these images was possible. The importance of
106
information about the fire observation is determined and an ideal fire observation
data types are given.
The ability to predict wildfire intensity, direction, rate of spread and
burned area is extremely important for wildfire management in terms of defining
suppression tactics managing financial and equipment resources for potential fires
and even active fires.
Fire model is essential for analyzing spatial fuel management activities and
examining suppression opportunities for fires that start at different locations or
under various weather scenarios. It also helps to determine the economic
consequences of potential fires with and without fuel management activities.
Lastly, it supports the strategic fire fighting decisions during fire or before fire.
Application of the GIS technology in modeling will decrease the
processing time and improves understanding the fire phenomena. Integration of
this technology to models is an inevitable part of fire simulation process since it
gives the ability of showing and analyzing all parts of the area in terms of fire
spread direction and rate, time of arrival, relative humidity.
In this study the suitability of a GIS – based forest fire simulation model,
namely FARSITE, is tested and the requirements of the model for Turkey are
determined. It should be remembered that fire simulation models can only
approximate reality. The output from a fire simulation system cannot replace the
107
knowledge and experience of wildland fire managers. Nevertheless, today’s fire
simulation systems are important tools that can help fire managers make better
decisions while saving time, money, and perhaps even lives.
5.1. Recommendation for Future Work
In order to improve the simulation accuracy of the fire model, some
recommendations have been listed.
• The most important parameter of the fire model is the description of the
fuel type. Fuel model of the study area must be defined initially. Even,
custom fuel models for Turkey can be defined by modeling the past fire or
studying test fires.
• During real forest fires, the information about the fireline position, the
suppression forces and its locations at a specific time, weather conditions
should be watched, measured in field and recorded. The ideal fire
observation data sheets and descriptions can be seen in Appendix I. By
using more detailed data, the fire behavior model can be calibrated
according to the real situations.
• The test fires should be studied for defining the record frequency of data
inputs in order to define the real situation.
108
• Aside from incorporating more sophisticated fire behavior models, other
dimensions of fire behavior and effects can be included. For example,
postfrontal combustion, spot fire.
• The fire simulation can be done by using various models. All these models
can be beneficial while obtaining fire spread area. But the important
parameters of these models are the data being used. In order to state the
appropriateness of the model, the same model must be validated for
another area.
• Especially, personnel that have experience about fire behavior gained from
real fire should work with the GIS specialists in order to define the fuel
model of the study area and calibrate the fuel model.
• Personnel of the General Directorate of Forest in Turkey should use fire
model for fire planning activities, for example, analyzing the effectiveness
of fuel treatments or checking the suppression opportunities and tactics for
various fires. The outputs of the simulation results can be used to maintain
strategic fire fighting decisions.
109
REFERENCES
Albini, F. A., 1976. Estimating wildfire behavior and effects. USDA For. Serv. Gen. Tech. Rep. INT-30.
Albini, F. A., 1979. Spotfire distance from burningtrees-a predictive model. USDA For. Serv. Gen. Tech. Rep. INT-56.
Albini, F. A. and Baughman, R. G., 1979. Estimating windspeeds for predicting wildland fire behavior. USDA For. Serv. Res. Pap. INT-221.
Albright D. and Meisner, B. N., 1999. Classification of Fire Simulation Systems. Fire Management Notes. Volume 59. No 2.
Anderson, H. E., 1982. Aids to Determining Fuel Models for Estimating Fire Behavior. USDA For. Serv. Gen. Tech. Rep. INT-122.
Andrews, P. L. 1986. BEHAVE: fire behavior prediction and fuel modeling system- BURN subsystem, Part 1. USDA For. Serv. Gen. Tech. Rep. INT-194.
Buckley, D.,2001. Integrating Remote Sensing and GIS Technologies to Support Fire Management. A Space Imaging White Paper.
Edmonds R. L., Agee, J. K., Gara R. I., 2000. Forest Health and Protection, McGraw-Hill
Fernandes, P., A., M., 2001. Fire spread prediction in shrub fuels in Portugal. Forest Ecology and Management 144:67 – 74. Elsevier.
110
Finney, M. A., 1998. FARSITE: Fire Area Simulator – Model Development and Evaluation. United States Department of Agriculutre Forest Service Research Paper RMRS-RP-4.
Finney, M. A. and Andrews, P. L., 1999. FARSITE – A Program for Fire Growth Simulation. Fire Management Notes, United States Department of Agriculture Forest Service, Vol. 59 No. 2: 13 – 15
Finney, M. A., 2003. FARSITE Version 4.0.3 Online Help.
Fujioka, F. M., 2001. Characterization and Analysis of Fire Spread Modeling Errors in an Integrated Weather/Wildland Fire Model. Doctor of Philosophy Thesis, University of California.
General Directory of Forest in Turkey, 2002. Evaluation Report of Fighting Forest Fires, Ankara, in Microsoft Excel media.
Hardwick, P., 1999. Study of Potential Benefits of Geographic Information Systems For Large Fire Incident Management. Prepared for: USDA Forest Service Region 5, Fire & Aviation Management.
Hawkes, B., Beck, J., and Sahle W., 1997. A Wildfire Threat Rating System For the McGregor Model Forest. Paper presented at the 13th Fire and Meteorology Conference, Lorne, Australia, Oct. 28 – 31, 1996.
Hawley, R. C. and Stickel, P. W., 1948. Forest Protection, Second Edition, John Wiley & Sons, New York, 351 p.
Jaiswal, R. K., Mukherjee, S., Raju, K. D., Saxena, R., 2002. Forest fire risk zone mapping from satellite imagery and GIS. International Journal of Applied Earth Observation and Geoinformation. Volume 4: 1-10.
Liu, P. S. and Chou, Y. H., 1997. A Grid Automation of Wildfire Growth Simulation. Paper presented at ESRI User Conference.
Marmaris – Çetibeli Forest Management Chief, 2002a. Damage report of Marmaris – Çetibeli Forest Fire, Information sent by Mr. Okan Turan via mail.
111
Marmaris – Çetibeli Forest Management Chief, 2002b. Evaluation report of Marmaris – Çetibeli Forest Fire, Information sent by Mr. Okan Turan via mail.
Noble, I. R., Bary, G. A. V., Gill, A. M., 1980. McArthur’s fire-danger meters expressed as equations. Australian Journal of Ecology 5: 201 – 203.
Noonan, E. K., 2002. The Application of FARSITE for assessing fuel treatments in the Eastern Sierra Nevada, Master Thesis, University of Nevada, Reno.
Rothermel, R. C., 1972. A mathematical model for predicting fire spread in wildland fuels. USDA For. Serv. Res. Pap. INT-115.
Rothermel, R. C., Rinehart, G. C., 1983. Field Procedures for Verification and Adjustment of Fire Behavior Predictions. United States Department of Agriculture Forest Service, General Technical Report INT – 142.
Rothermel, R. C., Wilson, R. A., Morris, G. A., and Sackett, S.S., 1986. Modeling moisture content of line dead wildland fuels input to the BEHAVE fire prediction system. USDA For. Serv. Res. Pap.INT-359.
Rothermel, R. C., 1991. Prediction Behavior and Size of Crown Fires in Northern Rocky Mountains. USDA For. Serv. Res. Pap. INT – 438.
The American Heritage® Dictionary of the English Language, Fourth Edition. Copyright © 2000 by Houghton Mifflin Company. Published by the Houghton Mifflin Company.
Van Wagtendonk, J. W., 1996. Use of a Deterministic Fire Growth Model to Test Fuel Treatments. Sierra Nevada Ecosystem Project, Final report to Congress. Vol II, Chap. 43: 1155-1166
Vega, J. A., Cuiňas, P., Fontúrbel, T., Pérez-Gorostiaga, P., Fernández, C., 1998. Predicting Fire Behavior in Galician (NW Spain) Shrubland Fuel Complexes.Departamento de Incendios Forestales. Centro de Investigaciones Forestales. Lourizán.
112
Weinstein, D., Green, K., Campbell, J., and Finney, M., 1995. Fire Growth Modeling in an integrated GIS Environment. Paper presented at ESRI User Conference
Internet References
Fire Area Simulator, FARSITE, Home page, http://farsite.org/ visited regularly
Fire Software : Information and Downloads, http://fire.org/ visited regularly
Web site of General Directorate of Forest in Turkey, http://www.ogm.gov.tr/
USDA Forest Service, http://www.fs.fed.us/
National Oceanic and Atmospheric Administration (NOAA) Web Site, http://www.noaa.gov/
113
APPENDIX A
NFFL FUEL MODEL DESCRIPTIONS
NFFL Fuel Model Description has been taken from the journal called as
“Aids to Determining Fuel Models For Estimating Fire Behavior” by Hal E
Anderson (1982).
GRASS GROUP
Fire Behavior Fuel Model 1
Fire spread is governed by the fine, very porous, and continuous
herbaceous fuels that have cured or are nearly cured. Fires are surface fires that
move rapidly through the cured grass and associated material. Very little shrub or
timber is present, generally less than one-third of the area. This fuel models
correlates to 1978 NFDRS fuel models A, L, S.
Fuel model values for estimating fire behavior
Total fuel load,< 3-inch dead and live, tons/acre 0.74
Dead fuel load, ¼ -inch, tons/acre .74
Live fuel load, foliage, tons/acre 0
Fuel bed depth, feet 1.0
114
Photo1. Western annual grasses such as cheatgrass, medusahead, ryegrass, and
rescues.
Photo 2. Live oak savanna of the South-west on the Coronado National Forest.
Photo 3. Open pine – grasslands on the Lewis and Clark National Forest
115
Fire Behavior Fuel Model 2
Fire spread is primarily through the fine herbaceous fuels either curing or
dead. These are surface fires where the herbaceous materials, in addition to litter
and dead-down stemwood form the open shrub or timber overstory, contribute to
the fire intensity. Open shrub lands and pine stands or scrub oak stands that cover
one-third to two-thirds of the area may generally fit this model, such stands may
include clumps of fuels that generate higher intensities and that may produce
firebrands. Some pinyon-juniper may be in this model. Photographs 4 and 5
illustrate possible field situations.
This fuel model correlates to 1978 NFDRS fuel models C and T.
Fuel model values for estimating fire behavior
Total fuel load,< 3-inch dead and live, tons/acre 4.0
Dead fuel load, ¼ -inch, tons/acre 2.0
Live fuel load, foliage, tons/acre 0.5
Fuel bed depth, feet 1.0
Photo 4. Open Ponderosa pine stand with annual grass understory
116
Photo 5. Scattered sage within grasslands on the Payette National Forest.
Fire Behavior Fuel Model 3
Fires in this fuel are the most intense of the grass group and display high
rates of spread under the influence of wind. Wind may drive fire into the upper
heights of the grass and across standing water. Stands are tall, averaging about 3
feet (1 m), but considerable variation may occur. Approximately one-third or
more of the stand is considered dead or cured and maintains the fire. Wild or
cultivated grains that have not been harvested can be considered similar to tall
prairie and marshland grasses. Refer to photographs 6, 7, and 8 for examples fo
fuels fitting this model.
This fuel correlates to 1978 NFDRS fuel model N.
Fuel model values for estimating fire behavior
Total fuel load,< 3-inch dead and live, tons/acre 3.0
Dead fuel load, ¼ -inch, tons/acre 3.0
Live fuel load, foliage, tons/acre 0
Fuel bed depth, feet 2.50
117
Fires in the grass group fuel models exhibit some of the faster rates of
spread under similar weather conditions. With a windspeed of 5 mi/h (8 km/h) and
a moisture content of 8 percent, representative rates of spread (ROS) are as
follows:
Model Rate of Spread
Chains/hour
Flame Length
Feet
1 78 4
2 35 6
3 104 12
As windspeed increases, model 1 will develop faster rates of spread than
model 3 due to fineness of the fuels, fuel load, and depth relations.
Photo 6. Fountaingrass in Hawaii; note the dead component.
118
Photo 7. Meadow foxtail in Oregon prairie and meadowland.
Photo 8. Sawgrass “prairie” and “strands” in the Everglades National Park, Fla
SHRUB GROUP
Fire Behavior Fuel Model 4
Fire intensity and fast spreading fires involve the foliage and live and dead
fine woody material in the crowns of a nearly continuous secondary overstory.
Stands of mature shrubs, 6 or more feet tall, such as California mixed chaparral,
the high pocosin along the east coast, the pinebarrens of New Jersey, or the closed
jack pine stands of the north-central States are typical candidates. Besides
flammable foliage, dead woody material in the stands significantly contributes to
the fire intensity. Height of stands qualifying for this model depends on local
119
conditions. A deep litter layer may also hamper suppression efforts. Photographs
9, 10, 11, and 12 depict examples fitting this fuel model.
This fuel model represents 1978 NFDRS fuel models B and O; fire
behavior estimates are more severe than obtained by models B or O.
Fuel model values for estimating fire behavior
Total fuel load,< 3-inch dead and live, tons/acre 13.0
Dead fuel load, ¼ -inch, tons/acre 5.0
Live fuel load, foliage, tons/acre 5.0
Fuel bed depth, feet 6.0
Photo 9. Mixed chaparral of southern California; note dead fuel component in
branchwood
120
Photo 10. Chaparral composed of manzanita and chamise near the Inaja Fire
Memorial, Calif.
Photo 11. Pocosin shrub field composed of species like fetterbush, gallberry, and
the bays.
Photo 12. High shrub southern rough with quantity of dead limbwood.
121
Fire Behavior Fuel Model 5
Fire is generally carried in the surface fuels that are made up of litter cast
by the shrubs and the grasses or forbs in the understory. The fires are generally
not very intense because surface fuel loads are light, the shrubs are young with
little dead material, and the foliage contains little volatile material. Usually shrubs
are short and almost totally cover the area. Young, green stands with no dead
wood would qualify: laurel, vine maple, alder, or even chaparral, manzanita, or
chamise.
No 1987 NFDRS fuel model is represented, but model 5 can be considered
as a second choice for NFDRS model D or as a third choice for NFDRS model T.
Photographs 13 and 14 show field examples of this type. Young green stands may
be up to 6 feet (2 m) high but have poor burning properties because of live
vegetation.
Fuel model values for estimating fire behavior
Total fuel load,< 3-inch dead and live, tons/acre 3.5
Dead fuel load, ¼ -inch, tons/acre 1.0
Live fuel load, foliage, tons/acre 2.0
Fuel bed depth, feet 2.0
Photo 13. Green, low shrub fields within timber stands or without overstory are
typical. Example is Douglas-fir-snowberry habitat type
122
Photo 14. Regeneration shrublands after fire or other disturbances have a large
green fuel component, Sundance Fire, Pack River Area, Idaho.
Fire Behavior Model 6
Fires carry through the shrub layer where the foliage is more flammable
than fuel model 5, but this requires moderate winds, greater than 8 mi/h (13 km/h)
at mid-flame height. Fire will drop to the ground at low wind speeds or at
openings in the stand. The shrubs are older, but not as tall as shrub types of model
4. A broad range of shrub conditions is covered by this model. Fuel situations to
be considered include intermediate stands of chamise, chaparral, oak brush, low
pocosin, Alaskan spruce talga, and shrub tundra. Even hardwood slash that has
cured can be considered. Pinyon-juniper shrublands may be represented but may
overpredict rate of spread except at high winds, like 20 mi/h (32 km/h) at 20-foot
level.
The 1978 NFDRS fuel models F and Q are represented by this fuel model.
It can be considered a second choice for models T and D and a third choice for
model S. Photographs 15, 16, 17, and 18 show situations encompassed by this fuel
model.
123
Fuel model values for estimating fire behavior
Total fuel load,< 3-inch dead and live, tons/acre 6.0
Dead fuel load, ¼ -inch, tons/acre 1.5
Live fuel load, foliage, tons/acre 0
Fuel bed depth, feet 2.5
Photo 15. Pinyon-juniper with sagebrush near Ely, Nev. understory mainly sage
with some grass intermixed.
Photo 16.Southern hardwood shrub with pine slash residues.
124
Photo 17. Low pocosin shrub field in the south.
Photo 18. Frost-killed Gambel Oak foliage, less than 4 feet in height, in Colorado
Fire Behavior Fuel Model 7
Fires burn through the surface and shrub strata with equal ease and can
occur at higher dead fuel moisture contents because of the flammability of live
foliage and other live material. Stands of shrubs are generally between 2 and 6
feet (0.6 and 1.8 m) high. Palmetto-gallberry understory-pine overstory sites are
typical and low pocosins may be represente. Black spruce-shrub combinations in
Alaska may also be represented.
125
The fuel model correlates with 1978 NFDRS model D and can be a second
choice for model Q. Photographs 19, 20, and 21 depict field situations for this
model.
Fuel model values for estimating fire behavior
Total fuel load,< 3-inch dead and live, tons/acre 4.9
Dead fuel load, ¼ -inch, tons/acre 1.1
Live fuel load, foliage, tons/acre 0.4
Fuel bed depth, feet 2.5
The shrub group of fuel models has a wide range of fire intensities and rate
of spread. With winds of 5 mi/h (8 km/h), fuel moisture content of 8 percent, and
a live fuel moisture content of 100 percent, the models have the values:
Model Rate of Spread
Chains/hour
Flame Length
Feet
4 75 19
5 18 4
6 32 6
7 20 5
Photo 19. Southern rough with light to moderate palmetto undertory.
126
Photo 20. Southern rough with moderate to heavy palmetto-gallberry and other
species.
Photo 21. Slash pine with gallberry, bay, and other species of understory rough
TIMBER GROUP
Fire Behavior Fuel Model 8
Slow-burning ground fires with low flame lengths are generally the case,
although the fire may encounter an occasional “jackpot” or heavy fuel
concentration that can flare up. Only under severe weather conditions involving
high temperatures, low humidities, and high winds do the fuels pose fire hazards.
Closed canopy stands of short-needle conifers or hardwoods that have leafed out
support fire in the compact litter layer. This layer is mainly litters, leaves, and
127
occasionally twigs because little undergrowth is present in the stand.
Representative conifer types are white pine, and lodgepole pine, spruce, fir, and
larch.
This model can be used for 1978 NFDRS fuel models H and R.
photographs 22, 23, and 24 illustrate the situations representative of this fuel.
Fuel model values for estimating fire behavior
Total fuel load,< 3-inch dead and live, tons/acre 5.0
Dead fuel load, ¼ -inch, tons/acre 1.5
Live fuel load, foliage, tons/acre 0
Fuel bed depth, feet 0.2
Photo 22. Surface litter fuels in western hemlock stands of Oregon and
Washington.
128
Photo 23. Understory of inland Dougles-fir has little fuel here to add to dead-
down litter load.
Photo 24. Closed stand of birch-aspen with leaf litter compacted.
Fire Behavior Fuel Model 9
Fires run through the surface litter faster than model 8 and have longer
flame height. Both long-needle conifer stands and hardwood stands, especially the
oak-hickory types, are typical. Fall fires in hardwoods are predictable, but high
winds will actually cause higher rates of spread than predicted because of spotting
caused by rolling and blowing leaves. Closed stands of long-needled pine like
ponderosa, Jeffrey, and red pines, or southern pine plantations are grouped in this
129
model. Concentrations of dead-down woody material will contribute to possible
torching out of trees, spotting, and crowning.
NFDRS fuel models E, P, and U are represented by this model. It is also a
second choice for models C and S. some of the possible field situations fitting this
model are shown in photographs 25, 26, and 27
Fuel model values for estimating fire behavior
Total fuel load,< 3-inch dead and live, tons/acre 3.5
Dead fuel load, ¼ -inch, tons/acre 2.9
Live fuel load, foliage, tons/acre 0
Fuel bed depth, feet 0.2
Photo 25. Western Oregon white oak fall litter; wind tumbled leaves may cause
short-range spotting that may increase ROS above the predicted value.
130
Photo 26. Loose hardwood litter under stands of oak, hickory, maple and other
hardwood species of the East.
Photo 27. Long-needle forest floor litter in ponderosa pine stand near Alberton,
Mont.
Fire Behavior Fuel Model 10
The fires burn in the surface and ground fuels with greater fire intensity
than the other timber litter models. Dead-down fuels include greater quantities of
3-inch (7.6-cm) or larger limbwood resulting from overmaturity or natural events
that create a large load of dead material on the forest floor. Crowning out,
spotting, and torching of individual trees are more frequent in this fuel situation,
leading to potential fire control difficulties. Any forest type may be considered if
heavy down material is present; examples are insect- or disease-ridden stands,
131
wind-thrown stands, overmature situations with deadfall, and aged light thinning
or partial-cut slash.
The 1978 NFDRS fuel model G is represented and is depicted in
photographs 28, 29, and 30.
Fuel model values for estimating fire behavior
Total fuel load,< 3-inch dead and live, tons/acre 12.0
Dead fuel load, ¼ -inch, tons/acre 3.0
Live fuel load, foliage, tons/acre 2.0
Fuel bed depth, feet 1.0
The fire intensities and spread rates of these timber litter fuel models are
indicated by the following values when the dead fuel moisture content is 8
percent, live fuel moisture content is 100 percent, and the effective windspeed at
midflame height is 5 mi/h (8 km/h);
Model Rate of Spread
Chains/hour
Flame Length
Feet
8 1.6 1.0
9 7.5 2.6
10 7.9 4.8
Fires such as above in model 10 are at the upper limit of control by direct
attack. More wind or drier conditions could lead to an escaped fire.
132
Photo 28. Old-growth Douglas-fir with heavy ground fuels.
Photo 29. Mixed conifer stand with dead-down woody fuels.
Photo 30. Spruce habitat type where succession or natural disturbance can
produce a heavy downed fuel load.
133
LOGGING SLASH GROUP
Fire Behavior Fuel Model 11
Fires are fairly active in the slash and herbaceous material intermixed with
the slash. The spacing of the rather light fuel load, shading from overstory, or the
aging of the fine fuels can contribute to limiting the fire potential. Light partial
cuts or thinning operations in mixed conifer stands, hardwood stands, and
southern pine harvests are considered. Clearcut operations generally produce more
slash than represented here. The less-than-3-inch (7.6-cm) material load is less
than 12 tones per acre (5.4 t/ha). The greater-than-3-inch (7.6-cm) is represented
by not more than 10 pieces, 4 inches (10.2 cm) in diameter, along a 50-foot (15 m)
transect.
The 1978 NFDRS fuel model K is represented by this model and field
examples are shown in photographs 31, 32, and 33.
Fuel model values for estimating fire behavior
Total fuel load,< 3-inch dead and live, tons/acre 11.5
Dead fuel load, ¼ -inch, tons/acre 1.5
Live fuel load, foliage, tons/acre 0
Fuel bed depth, feet 1.0
134
Photo 31. Slash residues left after sky-line logging in western Montana.
Photo 32. Mixed conifer partial cut slash residues may be similar to closed timber
with down woody fuels.
135
Photo 33. Light logging residues with patchy distribution seldom can develop
high intensities.
Fire Behavior Fuel Model 12
Rapidly spreading fires with high intensities capable of generating
firebrands can occur. When fire starts, it is generally sustained until a fuel break
or change in fuels is encountered. The visual impression is dominated by slash
and much of it is less than 3 inches (7.6 cm) in diameter. The fuels total less than
35 tones per acre (15.6 t/ha) and seem well distributed. Heavily thinned conifer
stands, clearcuts, and medium or heavy partial cuts are represented. The material
larger than 3 inches (7.6 cm) is represented by encountering 11 pieces, 6 inches
(15.2 cm) in diameter, along a 50-foot (15 m) transect.
This model depicts 1978 NFDRS model J and may overrate slash areas
when needles have dropped and the limbwood has settled. However, in areas
where limbwood breakup and general weathering have started, the fire potential
can increase. Field situations are presented in photographs 34, 35, and 36.
136
Fuel model values for estimating fire behavior
Total fuel load,< 3-inch dead and live, tons/acre 34.6
Dead fuel load, ¼ -inch, tons/acre 4.0
Live fuel load, foliage, tons/acre 0
Fuel bed depth, feet 2.3
Photo 34. Ponderosa pine clearcut east of Cascade mountain range in Oregon and
Washington
Photo 35. Cedar-hemlock partial cut in northern Idaho, Region 1, USFS
137
Photo 36. Lodgepole pine thinning slash on Lewis and Clark National Forest.
Red slash condition increases classification from light to medium.
Fire Behavior Fuel Model 13
Fire is generally carried across the area by a continuous layer of slash.
Large quantities of material larger than 3 inches (7.6 cm) are present. Fires spread
quickly through the fine fuels and intensity build up more slowly as the large fuels
start burning. Active flaming is sustained for long periods and a wide variety of
firebrands can be generated. These contribute to spotting problems as the weather
conditions become more severe. Clearcuts and heavy partial-cuts in mature and
overmature stands are depicted where the slash load is dominated by the greater
than 3 inches (7.6-cm) diameter in material. The total load may exceed 200 tones
per acre (89 t/ha) but fuel less than 3 inches (7.6 cm) is generally only 10 percent
of the total load. Situations where the slash still has “red” needles attached but the
total load is lighter, more like model 12, can be represented because of the earlier
high intensity and quicker area involvement.
The 1978 NFDRS fuel model I is represented and is illustrated in
photographs 37 and 38. Areas most commonly fitting this model are old-growth
stands west of the Cascade and Sierra Nevada Mountains. More efficient
utilization standards are decreasing the amount of large material left in the field.
138
Fuel model values for estimating fire behavior
Total fuel load,< 3-inch dead and live, tons/acre 58.1
Dead fuel load, ¼ -inch, tons/acre 7.0
Live fuel load, foliage, tons/acre 0
Fuel bed depth, feet 3.0
For other situations:
Hardwood slash ………………………………………………… Model 6
Heavy “red” slash ………………………………………………. Model 4
Overgrown slash ………………………………………………... Model 10
Southern pine clearcut slash ……………………………………. Model 12
The comparative rates of spread and flame lengths for the slash models at
8 percent dead fuel moisture content and a 5 mi/h (8 km/h) midflame wind are:
Model Rate of Spread
Chains/hour
Flame Length
Feet
11 6.0 3.5
12 13.0 8.0
13 13.5 10.5
Photo 37. West coast Douglas-fir clearcut, quantity of cull high
139
Photo 38. High productivity of cedar-fir stand can result in large quantities of
slash with high fire potential.
140
APPENDIX B
PARAMETERS AND EXAMPLES OF FARSITE ASCII
INPUT FILES
i. Weather (.WTR)
Month Day Precip Hour1 Hour2 Temp1 Temp2 Humid1 Humid2 Elevation rt1 rt2 • Precipitation is the daily rain amount specified in hundredths of an inch or
millimeters (integer). • Hour1 corresponds to the hour at which the minimum temperature was
recorded (0-2400). • Hour2 corresponds to the hour at which the maximum temperature was
recorded (0-2400). • Temperatures (Temp1 is minimum; Temp2 is maximum) are in degrees
Fahrenheit or Celsius (integer). • Humidities (Humid1 is maximum; Humid2 is minimum) are in percent, 0
to 99 (integer). • Elevation is in feet or meters above sea level. NOTE: these units (feet or
meters) do not have to be the same as the landscape elevation theme (integer).
• Precipitation Duration is entered with the beginning (rt1) and ending (rt2)
times (0-2400) of the daily rain amount. Only one time period per day is allowed.
141
Examples
ENGLISH8 10 00 600 1500 48 99 94 11 24008 11 00 600 1500 46 96 78 14 24008 12 07 600 1600 48 90 74 14 2400 1830 2000
ii. Wind (.WND)
Month Day Hour Speed Direction CloudCover
• Hour is specified as 0-2400, to the nearest minute (integer). • Speed is either the 20ft windspeed specified in miles per hour or the 10m
windspeed in kilometers per hour (integer) • Direction is specified in degrees, clockwise from north (0-360), (integer).
A "-1" in the direction field indicates the winds to be up slope, similarly downslope winds can be specified with a "-2".
• CloudCover is specified as a percentage, 0 to 100 (integer).
Example
ENGLISH8 10 0 1 54 08 10 100 2 67 08 10 200 2 102 08 10 300 1 166 08 10 400 3 319 0
iii. Adjustment Factors (.ADJ)
FuelMod AdjustmentFactor
• FuelMod is an integer value (1-50). As always, model numbers 1-13 are restricted to the 13 standard Fire Behavior fuel models (Anderson 1982). Models 14-50 are for custom models specified in the Fuel Model (.FMD) File.
• The AdjustmentFactor can be a floating point number (decimal)
specifying the multiplier for rate of spread adjustment (see above). It must be greater than zero.
142
Example
1 .52 .753 .56 .57 .510 .7511 .512 .513 .5
iv. Initial Fuel Moisture (.FMS)
FuelMod 1Hour 10Hour 100Hour LiveH LiveW
• Fuel Model (1-50) corresponds to a fuel model specified 1. on the landscape if no conversions are used, or 2. in the Fuel Conversion (.FMS) File if conversions are used. Fuel
models 1-13 must relate to the Fire Behavior 13 standard fuel models (Anderson 1982). Fuel models numbered from 14 to 50 are for custom models as described in the Custom Fuel Model (.FMD) File.
• Fuel moistures for each category are in percent (integers), and may
exceed 100. LiveH and LiveW indicate "live woody" and "live herbaceous" fuels, just like BehavePlus. Unlike dead fuels, live fuel moistures remain constant throughout the simulation unless you manually change them
Example
1 3 4 6 50 752 3 4 6 50 753 3 4 6 50 754 3 4 6 50 755 3 4 6 75 1006 3 4 6 50 1007 3 4 6 50 758 4 5 7 75 1009 3 4 5 50 7510 4 5 7 75 10011 3 4 6 50 7512 4 5 7 75 10013 4 5 7 75 100
143
v. Fuel Model Conversion (.CNV)
ConvertFrom FuelMod
• ConvertFrom is the index or attribute of a fuel type in the fuels theme converted to a fuel model number 1-99, (integer)
• FuelMod can be any fuel model 1-50, (integer)
Example 68 274 586 85 68 9
vi. Custom Fuel Models (.FMD)
FMod 1H 10H 100H LiveH LiveW 1HSAV LiveHSAV LiveWSAV Depth XtMoist DHt LHt
Field Name Data Type English Units Metric Units FMod Fuel Model integer number 14-50 number 14-50 1H, 10H, 100H, LiveH, LiveW
Fuel Loading decimal tons/acre metric tonnes/hectare
1HSAV, LiveHSAV, LiveWSAV
Surface to Volume Ratio
integer 1/ft 1/cm
Depth Fuel Bed Depth
decimal Ft cm
XtMoist Moisture of Extinction
integer percent percent
DHt, LHt Heat Content, live & dead fuels
integer BTU/lb J/Kg
Example ENGLISH
19 2.250 1.500 3.710 0.000 1.000 2000 1800 1500 0.600 25 8000 8000
144
vii. Air Attack Resources (.AIR)
#name of 1st aircraft# bracket the “name” with # Units METERS or FEET 1 line_length Coverage level 1, line length a decimal number 2 line_length Coverage level 2, line length a decimal number 3 line_length Coverage level 3, line length a decimal number 4 line_length Coverage level 4, line length a decimal number 6 line_length Coverage level 6, line length a decimal number 8 line_length Coverage level 8, line length a decimal number Return_Time Return time in minutes COST_PER_HOUR 0.00 Optional input #Name of 2nd aircraft# Append other aircraft descriptions etc.
Example
#C-130 3000 gal#FEET1 3002 1503 1004 756 508 40RETURN_TIME 120COST 6000
viii. Coarse Woody Profile (.CWD)
SizeClass Loading HeatContent S/R Moist
• SizeClass - The representative size of the class based on surface to volume ratio. (i.e.; for the 3" to 6" size class the representative size is 4.75, for the 6" to 10" class it is 8.25). A decimal data type, the units are inches or centimeters.
• Loading - Fuel loading of the class (decimal), units are tons/acre or
kilograms/hectare • HeatContent - Heat content of the class (integer), units are BTU/lb or
joules/kilogram • S/R - Sound or rotten is defined by the density of the fuel (lb/ft3, or
kg/m3). Typical values are 32 lb/ft3 for sound fuel and 19 lb/ft3 for rotten.
145
• Moist - Moisture content of the size class in percent (integer).
Example ENGLISHMODEL 8 CWD_for_fm_8DEPTH 1.640.024 1.50 8000 32 40.440 1.00 8000 32 51.600 2.50 8000 32 74.750 1.30 7997 32 11MODEL 9 CWD_for_fm_9DEPTH 0.500.019 2.92 8000 32 20.440 0.41 8000 32 41.600 0.15 8000 32 54.750 0.60 7997 32 10
ix. Burn Period (.BPD)
Month Day StartHour EndHour
Example 8 9 0 24008 10 0 24008 11 800 24008 12 800 2400
x. Gridded Weather File (.ATM)
The format of both weather and wind variables in gridded is:
WEATHER_AND_WINDS MONTH DAY HOUR TEMP HUMID PPT WSPEED WDIR CLDCOVER
Example WEATHER_AND_WINDS7 15 0200 TEST01.TMP TEST01.HMD TEST01.PPT TEST01.SPD TEST01.DIRTEST01.CLD7 15 0600 TEST02.TMP TEST02.HMD TEST02.PPT TEST02.SPD TEST02.DIRTEST02.CLD7 15 1200 TEST03.TMP TEST03.HMD TEST03.PPT TEST03.SPD TEST03.DIRTEST03.CLD7 15 1600 TEST04.TMP TEST04.HMD TEST04.PPT TEST04.SPD TEST04.DIRTEST04.CLD7 15 2000 TEST05.TMP TEST05.HMD TEST05.PPT TEST05.SPD TEST05.DIRTEST05.CLD
146
7 16 0200 TEST06.TMP TEST06.HMD TEST06.PPT TEST06.SPD TEST06.DIRTEST06.CLD7 16 0600 TEST07.TMP TEST07.HMD TEST07.PPT TEST07.SPD TEST07.DIRTEST07.CLD7 16 1200 TEST08.TMP TEST08.HMD TEST08.PPT TEST08.SPD TEST08.DIRTEST08.CLD7 16 1600 TEST09.TMP TEST09.HMD TEST09.PPT TEST09.SPD TEST09.DIRTEST09.CLD7 16 2000 TEST10.TMP TEST10.HMD TEST10.PPT TEST10.SPD TEST10.DIRTEST10.CLD
The format of only wind variables in gridded format is
WINDS_AND_CLOUDS MONTH DAY HOUR WSPEED WDIR CLDCOVER
Example
WINDS_AND_CLOUDS7 15 0200 TEST01.SPD TEST01.DIR TEST01.CLD7 15 0600 TEST02.SPD TEST02.DIR TEST02.CLD7 15 1200 TEST03.SPD TEST03.DIR TEST03.CLD7 15 1600 TEST04.SPD TEST04.DIR TEST04.CLD7 15 2000 TEST05.SPD TEST05.DIR TEST05.CLD7 16 0200 TEST06.SPD TEST06.DIR TEST06.CLD7 16 0600 TEST07.SPD TEST07.DIR TEST07.CLD7 16 1200 TEST08.SPD TEST08.DIR TEST08.CLD7 16 1600 TEST09.SPD TEST09.DIR TEST09.CLD7 16 2000 TEST10.SPD TEST10.DIR TEST10.CLD
xi. Ground Attack Resources
#name of 1st crew# bracket the "name" with # units METERS_PER_MINUTE, or
FEET_PER_MINUTE, or CHAINS_PER_HOUR
FLAME_LIMIT 0.0 meters or feet depending on units designation above
fuel_model line_production_rate fuel model is an integer, production rate a decimal number
fuel_model line_production_rate 99 last fuel model line, no production rate COST_PER_HOUR 0.00 optional input #Name of 2nd Crew# Append other crew descriptions etc.
147
Example #T-Falls Engine#CHAINS_PER_HOURFLAME_LIMIT 6.0000001 5.002 5.003 2.984 2.985 1.506 2.987 2.988 4.009 3.5010 2.9811 2.9812 2.0013 1.5014 2.9815 2.9899COST_PER_HOUR 85.00#T-Falls Hand Crew#CHAINS_PER_HOURFLAME_LIMIT 3.0000001 2.982 2.983 2.98
148
APPENDIX C
FIRE REGISTER REPORT
149
150
APPENDIX D
LINEAGE REPORT
Projection Parameters Universal Transverse Mercator European Datum 1950 Zone 35N (UTM ED 50 Zone 35N (International Spheroid 1909)) Projection: Transverse_MercatorParameters:
False_Easting: 500000.000000False_Northing: 0.000000Central_Meridian: 27.000000Scale_Factor: 0.999600Latitude_Of_Origin: 0.000000
Linear Unit: Meter (1.000000)Geographic Coordinate System:Name: GCS_European_1950Angular Unit: Degree (0.017453292519943295)Prime Meridian: Greenwich (0.000000000000000000)Datum: D_European_1950
Spheroid: International_1924Semimajor Axis: 6378388.000000000000000000Semiminor Axis: 6356911.946127946500000000Inverse Flattening: 297.000000000000000000
Aydın N 20 c3 link table1.634650 0.864550 623000.000000 4096000.0000001.960928 21.300232 623000.000000 4109000.00000017.589541 21.056193 633000.000000 4109000.00000017.270546 0.641050 633000.000000 4096000.0000009.455173 0.757264 628000.000000 4096000.0000009.773692 21.177405 628000.000000 4109000.0000009.691316 16.469064 628000.000000 4106000.0000009.540174 7.030185 628000.000000 4100000.0000001.877058 16.592877 623000.000000 4106000.0000009.612459 11.742522 628000.000000 4103000.00000017.508491 16.346854 633000.000000 4106000.00000017.429396 11.624916 633000.000000 4103000.000000
151
17.359346 6.913005 633000.000000 4100000.0000001.723297 7.143457 623000.000000 4100000.0000001.797052 11.863168 623000.000000 4103000.0000005.081246 21.253460 625000.000000 4109000.0000005.000231 16.540619 625000.000000 4106000.0000004.921364 11.815783 625000.000000 4103000.0000004.847440 7.098215 625000.000000 4100000.0000004.761110 0.825419 625000.000000 4096000.00000014.457851 21.104982 631000.000000 4109000.00000014.376944 16.395997 631000.000000 4106000.00000014.301435 11.672706 631000.000000 4103000.00000014.227821 6.958488 631000.000000 4100000.00000014.142009 0.688322 631000.000000 4096000.0000001.675670 4.008182 623000.000000 4098000.0000004.802305 3.962153 625000.000000 4098000.0000009.496978 3.896749 628000.000000 4098000.00000014.183354 3.827488 631000.000000 4098000.00000017.312537 3.779122 633000.000000 4098000.000000 Transformation: 3rd order polynomial Total RMS error: 5.05341 Rectification: Cell Size: 3.240321 Resample Type: Bilinear Interpolation (for continuous data) Aydın N 20 c4 link table1.772453 1.128789 612000.000000 4096000.00000017.729381 21.309283 622000.000000 4109000.00000017.400277 0.896574 622000.000000 4096000.0000002.102962 21.556195 612000.000000 4109000.0000009.749527 12.003378 617000.000000 4103000.0000001.832468 5.835604 612000.000000 4099000.0000001.907764 10.553003 612000.000000 4102000.0000009.587899 1.016089 617000.000000 4096000.00000017.564921 11.882279 622000.000000 4103000.0000009.913591 21.428875 617000.000000 4109000.0000009.677290 7.287932 617000.000000 4100000.0000009.830800 16.726710 617000.000000 4106000.00000017.489395 7.167833 622000.000000 4100000.00000017.646470 16.602137 622000.000000 4106000.0000002.013488 16.850902 612000.000000 4106000.0000005.221392 21.506123 614000.000000 4109000.0000005.137576 16.796465 614000.000000 4106000.0000005.058404 12.076236 614000.000000 4103000.0000004.982593 7.358601 614000.000000 4100000.0000004.917522 2.655558 614000.000000 4097000.00000014.598184 21.358432 620000.000000 4109000.00000014.516432 16.652304 620000.000000 4106000.00000014.434168 11.932102 620000.000000 4103000.00000014.361588 7.217755 620000.000000 4100000.00000014.293106 2.514407 620000.000000 4097000.0000009.607990 2.587400 617000.000000 4097000.0000001.791030 2.697138 612000.000000 4097000.000000
152
17.423957 2.466320 622000.000000 4097000.0000001.857558 7.404282 612000.000000 4100000.0000001.931660 12.126461 612000.000000 4103000.000000
Transformation: 3rd order polynomial Total RMS error: 5.11556 Rectification: Cell Size: 3.241644 Resample Type: Bilinear Interpolation (for continuous data) Marmaris O 20 b1 link table1.896362 21.429171 612000.000000 4095000.00000017.193754 0.759066 622000.000000 4082000.00000017.524501 21.186032 622000.000000 4095000.0000001.571911 1.004197 612000.000000 4082000.0000009.711217 21.309665 617000.000000 4095000.00000012.838356 21.260757 619000.000000 4095000.0000006.591083 21.358272 615000.000000 4095000.0000001.821600 16.724312 612000.000000 4092000.0000001.741816 12.003283 612000.000000 4089000.0000001.668934 7.273865 612000.000000 4086000.0000001.595427 2.566307 612000.000000 4083000.0000006.511907 16.652596 615000.000000 4092000.0000006.434125 11.927548 615000.000000 4089000.0000006.355805 7.201642 615000.000000 4086000.0000006.282494 2.494486 615000.000000 4083000.0000009.632729 16.605006 617000.000000 4092000.0000009.552823 11.877379 617000.000000 4089000.0000009.478828 7.152753 617000.000000 4086000.0000009.403873 2.446207 617000.000000 4083000.00000012.757323 16.557545 619000.000000 4092000.00000012.682603 11.827731 619000.000000 4089000.00000012.606602 7.103264 619000.000000 4086000.00000012.535359 2.397598 619000.000000 4083000.00000017.445005 16.484200 622000.000000 4092000.00000017.367424 11.752786 622000.000000 4089000.00000017.288271 7.033196 622000.000000 4086000.00000017.216716 2.325737 622000.000000 4083000.0000006.257964 0.928608 615000.000000 4082000.0000009.379728 0.879948 617000.000000 4082000.00000012.512181 0.831497 619000.000000 4082000.000000
Transformation: 3rd order polynomial Total RMS error: 4.47960 Rectification: Cell Size: 3.242143 Resample Type: Bilinear Interpolation (for continuous data)
153
Marmaris O 20 b2 link table1.646805 21.025619 623000.000000 4095000.00000016.935584 1.920207 633000.000000 4083000.00000017.256756 20.762152 633000.000000 4095000.0000001.302053 0.597927 623000.000000 4082000.0000009.458683 20.894843 628000.000000 4095000.0000009.373207 16.187377 628000.000000 4092000.0000009.291548 11.465809 628000.000000 4089000.0000009.207934 6.751826 628000.000000 4086000.0000009.133969 2.047194 628000.000000 4083000.00000017.174447 16.058131 633000.000000 4092000.00000017.090986 11.342838 633000.000000 4089000.00000017.012318 6.627345 633000.000000 4086000.0000001.562389 16.313885 623000.000000 4092000.0000001.483438 11.595352 623000.000000 4089000.0000001.402385 6.874850 623000.000000 4086000.0000006.331901 20.948637 626000.000000 4095000.0000006.248448 16.238598 626000.000000 4092000.0000006.166970 11.518297 626000.000000 4089000.0000006.089065 6.799260 626000.000000 4086000.0000006.015148 2.097593 626000.000000 4083000.00000012.574356 20.841992 630000.000000 4095000.00000012.491779 16.133889 630000.000000 4092000.00000012.411016 11.413024 630000.000000 4089000.00000012.331839 6.701476 630000.000000 4086000.00000012.257021 1.994799 630000.000000 4083000.0000001.328561 2.172649 623000.000000 4083000.0000006.220615 14.662159 626000.000000 4091000.00000012.464279 14.557675 630000.000000 4091000.00000012.357725 8.268915 630000.000000 4087000.0000006.112636 8.373853 626000.000000 4087000.000000
Transformation: 3rd order polynomial Total RMS error: 3.19300 Rectification: Cell Size: 3.247341 Resample Type: Bilinear Interpolation (for continuous data) Transformation: 3rd order polynomial Total RMS error: 4.41897 Cetibeli_mescere_cut link table2952.029220 -410.429349 618791.270271 4097572.7884791161.025634 -511.996115 615949.068459 4097409.4527711240.406900 -2969.508462 616062.203930 4093487.7610522778.508318 -2953.365809 618490.635286 4093517.4361803284.000545 -3570.105985 619280.377414 4092540.1142613737.887205 -2950.911729 620000.786759 4093522.5538425151.940884 -3228.631641 622233.528497 4093081.7019927245.403046 -3263.151541 625535.215515 4093029.145432
154
4201.618498 -2405.221603 620738.257341 4094395.3692593338.202273 -1069.942098 619381.548596 4096521.9205797531.363806 -765.839516 626002.421584 4096999.2600358167.549080 -2653.562364 627002.096983 4094001.0583926265.792991 -131.782739 624004.478252 4098001.7022682573.783017 -328.130204 618185.786596 4097694.2489474624.617482 -1251.209777 621403.729024 4096233.1248012401.696228 -1800.335044 617892.983549 4095358.088597
Transformation: 3rd order polynomial Total RMS error: 5.58809 Create TIN from Features Parameters Layers: Contour Settings for Selected Layer Feature Type: 3D Lines Height Source: <Feature Z Values> Triangulate as: Hard line Tag value field: None Interpolate to Raster Kriging Parameters Input points: Elevation.shp Z value field: Elevation Kriging Method: Ordinary Semivariogram model: Spherical Search radius type: Fixed Search Radius Settings Distance 100.000 Minimum Number of Points: 0 Output cell size: 20.000
Table D.1. Comparison of 10 Elevation Data Which Were Read From DEM and Topographic Map Number Coordinate X
(m) Coordinate Y
(m) Contour Value
(m) DEM Value
(m) 1 617904.731130 4095380.265906 205 206.2699 2 622325.261711 4093897.986090 380 379.9669 3 622465.638234 4093989.647473 450 449.0404 4 624206.709838 4093793.574605 160 160.123 5 618183.743189 4093778.706455 330 330.4601 6 620244.173474 4093578.007321 530 529.9805 7 620366.047750 4096781.060286 60 59.9881 8 619111.804242 4096855.062015 200 199.6599 9 622849.492920 4095734.000660 100 100.2068 10 622497.987262 4092843.071980 440 440.4774
155
Table D.2. Lineage Information of Digital Maps File
Name Type Source Date Method of
Production Map Scale
Projection
Spheroid
Datum
Rectifyaydin-n20-c3.tif
Raster Paper Map
1996 Scanned, registered by ArcGIS 8.1 and rectified.
1:25,000 UTM Zone 35
Int 1909 ED50
Rectifyaydin-n20-c4.tif
Raster Paper Map
1996 Scanned, registered by ArcGIS 8.1 and rectified.
1:25,000 UTM Zone 35
Int 1909 ED50
Rectifymarmaris-o20-b1.tif
Raster Paper Map
1996 Scanned, registered by ArcGIS 8.1 and rectified.
1:25,000 UTM Zone 35
Int 1909 ED50
Rectifymarmaris-o20-b2.tif
Raster Paper Map
1996 Scanned, registered by ArcGIS 8.1 and rectified.
1:25,000 UTM Zone 35
Int 1909 ED50
River.shp Vector Polyline
Raster Maps
5/24/2003
Digitized by using ArcGIS 8.1
UTM Zone 35
Int 1909 ED50
Road.shp Vector Polyline
Raster Maps
5/24/2003
Digitized by using ArcGIS 8.1
UTM Zone 35
Int 1909 ED50
contour.shp
Vector Polyline ZM
Raster Maps
5/23/2003
Digitized by using ArcGIS 8.1
UTM Zone 35
Int 1909 ED50
elevation.shp
Vector PointZM
contour.shp
5/23/2003
Generated from TIN and joined with contour.shpbased on spatial location.
UTM Zone 35
Int 1909 ED50
marmaris20
Raster DEM
elevation.shp
Generated by Kriging Interpolation Method
UTM Zone 35
Int 1909 ED50
cetibelimescere.ecw
Raster Paper Maps
Scanning and registered by collecting reference points from topographic maps
1:25,000 - - -
mescere.shp
Vector Polyline
Cetibelimescere.ecw
Digitized by using ArcGIS 8.1
1:25,000 UTM Zone 35
Int 1909 ED50
cet00.tif Raster IKONOS Reprojected, stacked and merged with panchromatic image by ERDAS Imagine
UTM Zone 35
Int 1909 ED50
cet01.tif Raster IKONOS Reprojected, stacked and merged with panchromatic image by ERDAS Imagine
UTM Zone 35
Int 1909 ED50
156
APPENDIX E
TABLE OF VEGETATION SYMBOLS
Table E.1. Table of Vegetation Symbols Use by General Directorate of Forest İBRELİLER YAPRAKLILAR YAPRAKLILAR Kot No
Sem-bolü
Ağaç Türü Kot No
Sem-bolü
Ağaç Türü Kot No
Sem-bolü
Ağaç Türü
01 Çz Kızılçam 21 Kn Kayın 41 Cv Ceviz 02 Çk Karaçam 22 M Meşe 42 Zy Yabanizeytin 03 Çs Sarıçam 23 Gn Gürgen 43 Mp Palamutmeşesi 04 G Göknar 24 Kz Kızılağaç 44 Ms Saplımeşe 05 L Ladin 25 Kv Kavak 45 Mz Sapsızmeşe 06 S Sedir 26 Ks Kestane 46 Mc Macarmeşesi 07 At Ardıç 27 Dş Dişbudak 47 Mt Tüylümeşe 08 Cf Fıstıkçamı 28 Ih Ihlamur 48 Mm Mazımeşesi 09 Sr Servi 29 Ak Akçaağaç 49 Ml Saçlımeşe 10 P Porsuk 30 Ka Karaağaç 50 Mr Pırnalmeşesi 11 Çh Halepçamı 31 Ky Kayacık 51 Mk Kermezmeşesi 12 Çm Sahilçamı 32 Çn Çınar 52 Ko Kocayemiş 13 Çr P. Radiata 33 Ok Okaliptüs 53 Ma Maki 14 D Duglaz 34 Sğ Sığla 54 15 An Andız 35 Fn Fındık 55 16 36 Sö Söğüt 56 17 37 H Huş 57 18 38 Df Defne 58 19 39 Ş Şimşir 59 20 Di Diğer İbreli 40 O Ormangülü 60 Dy Diğer Yapraklı
Yukarıdaki tabloda bulunmayan ağaç türleri; ilk harfleri gözönünde bulundurulmak
suretiyle sembolleştirilerek ibrelilerde (16 – 19), yapraklılarda (54 – 59) kod numaraları kullanılır. Ancak bu kod numaraları da yetişmediği taktirde; ibrelilerde (Di sembolü) ve (20) kod numarası, yapraklılarda (Dy) sembolü ve (60) kod numarası kullanılır.
Korularda: Örnek :Çz0 : Prodüktif kızılçam boşaltılmış gençleştirme alanı Çz0Y : Prodüktif kızılçam boşaltılmış yanık alanı,
Çzc2Y : prodüktif kızılçam boşaltılmamış “ c “ çağlı iki kapalı yanık mesceresi.
157
Table E.1. Table of Vegetation Symbols Use by General Directorate of Forest
(Continued)
Bozuk korularda : Mescere tipleri esas ağaç türünün başına B harfleri konulmak süreti ile sembolleştirilir.
Örnek :BÇz : Bozuk kızılçam mesceresi, BÇz0Y : Bozuk kızılçam boşaltılmış yanık alanı, BÇzY : Bozuk kızılçam boşaltılmamış yanık mesceresi. Baltalıklarda : Baltalık kelimesi Bt rumuzu ile gösterilir. Bozuk baltalıkta ise B
harfleri konulmak suretiyle sembolleştirilir. Baltalık mescere tipi sembollerinin yazılmasında; karışık ağaç türleri K harfi ile gösterilir.
Örnek :MBt2/00 = Meşe baltalığı, 2 kapalı, yeni kesilmiş saha. MBt3/05 = Meşe baltalığı, 3 kapalı, 5 yaşında. MBt2/10 = Meşe Baltalığı, 2 kapalı, 10 yaşında MBt1/15 = Meşe Baltalığı, 1 kapalı, 15 yaşında, KBt3/01 = Karışık baltalık, 3 kapalı, 1 yaşında. KBt2/08 = Karışık baltalık, 2 kapalı, 8 yaşında. KBt1/27 = Karışık baltalık, 1 kapalı, 27 yaşında. BMBt =Bozuk meşe baltalığı. BKBt = Bozuk karışık baltalık. Ağaç türleri karışık korularda semboller: GL = Göknar Ladin, GKnL = Göknar Kayın Ladin Mescere tipinde, iki gelişme çağınında yanyana gösterilmesinin zorunlu olduğu
hallerde; hakim gelişme çağı rumuzu önce yazılır. Örnek : bc = b çağ sınıfı hakim.
cd = c çağ sınıfı hakim. Ormansız sahaların sembolleri : OT : Ağaçsız orman toprağı. E : Erozyonlu saha. F : Orman fidanlığı. T : Kayalık, Taşlık. Ku : Kum. Bk : Bataklık, Sazlık. Su : Göl, Bent, Baraj, Nehir. Me : Mer’a, Otlak, Yayla, Çayır, Bozkır. İs : İskan sahası, Mezarlık. Dp : Orman deposu ve istif yeri. Z : Tarım arazisi (Tarla, Meyvelik, Sebzelik, Bağlık v.s. gibi)
158
APPENDIX F
RAW METEOROLOGICAL DATA
The meteorological data, which are hourly air temperature, humidity and
cloud cover, and hourly wind speed and direction, have been obtained from
Turkish State Meteorological Service.
159
160
161
162
APPENDIX G
ASCII INPUT FILES OF ÇETİBELİ
Initial ASCII Files
Weather Data File (.WTR)
METRIC8 14 0 600 1400 24 34 71 42 168 15 0 600 1200 25 33 67 35 168 16 0 500 1400 23 32 63 39 168 17 0 600 1400 24 31 70 42 168 18 0 600 1300 23 32 83 41 168 19 0 600 1400 24 32 82 41 168 20 0 600 1600 23 31 87 53 168 21 0 500 1300 23 32 91 42 168 22 0 600 1400 23 32 73 38 168 23 0 300 1500 24 33 78 37 168 24 0 500 1400 24 34 82 39 168 25 0 600 1400 24 35 94 38 168 26 0 600 1300 25 35 74 35 168 27 0 600 1400 25 34 79 38 16
Wind Data File (.WND)
METRIC8 14 0 7 293 08 14 100 8 293 08 14 200 6 293 08 14 300 7 293 08 14 400 8 293 08 14 500 7 315 08 14 600 5 293 08 14 700 6 270 08 14 800 7 270 08 14 900 10 270 0
163
8 14 1000 11 270 08 14 1100 16 270 08 14 1200 18 270 08 14 1300 17 270 08 14 1400 19 270 08 14 1500 19 270 08 14 1515 49 293 08 14 1600 14 293 08 14 1700 14 270 08 14 1800 10 293 08 14 1900 9 293 08 14 2000 8 293 08 14 2100 11 293 08 14 2200 8 293 08 14 2300 15 293 08 15 0 14 293 08 15 100 6 293 08 15 200 6 293 08 15 300 10 293 08 15 400 12 293 08 15 500 12 293 08 15 600 11 293 08 15 700 9 293 08 15 800 8 270 08 15 900 10 338 08 15 1000 11 338 08 15 1100 12 293 08 15 1200 16 248 08 15 1300 22 293 08 15 1400 22 293 08 15 1434 50 293 08 15 1500 22 293 08 15 1600 21 270 08 15 1700 16 270 08 15 1800 13 270 08 15 1900 12 293 08 15 2000 12 293 08 15 2100 10 293 08 15 2200 10 293 08 15 2300 14 293 08 16 0 8 293 08 16 100 6 293 08 16 200 9 293 08 16 300 6 293 08 16 400 9 293 08 16 500 9 293 08 16 600 11 293 08 16 700 13 293 08 16 800 7 293 08 16 900 9 270 08 16 1000 14 270 08 16 1100 15 293 08 16 1200 15 293 08 16 1300 15 270 08 16 1400 14 270 08 16 1500 16 270 08 16 1536 42 270 08 16 1600 15 270 08 16 1700 18 293 0
164
8 16 1800 15 293 08 16 1900 16 293 08 16 2000 14 293 08 16 2100 12 293 08 16 2200 9 293 08 16 2300 8 293 0
Adjustment File (.ADJ)
1 1.0000002 1.0000003 1.0000004 1.0000005 1.0000006 1.0000007 1.0000008 1.0000009 1.00000010 1.00000011 1.00000012 1.00000013 1.000000
Initial Fuel Moisture (.FMS)
1 3 4 6 50 752 3 4 6 50 753 3 4 6 50 754 3 4 6 50 755 3 4 6 75 1006 3 4 6 50 1007 3 4 6 50 758 4 5 7 75 1009 3 4 5 50 7510 4 5 7 75 10011 3 4 6 50 7512 4 5 7 75 10013 4 5 7 75 100
Modified ASCII Input Files
Custom Fuel Models (.FMD)
METRIC14 0.500 0.000 0.000 0.000 0.000 110 54 41 9.140 12 1859215 25.650 0.000 0.000 0.000 0.000 47 58 47 289.560 25 1859216 24.890 3.500 1.290 0.000 0.000 78 54 41 23.180 25 18592
165
Initial Fuel Moisture (.FMS) File
1 3 4 6 50 752 3 4 6 50 753 3 4 6 50 754 3 4 6 50 755 3 4 6 75 1006 3 4 6 50 1007 3 4 6 50 758 4 5 7 75 1009 3 4 5 50 7510 4 5 7 75 10011 3 4 6 50 7512 4 5 7 75 10013 4 5 7 75 10014 3 4 6 50 7515 3 4 6 50 7516 3 4 5 50 75
Fuel Model Conversion (.CNV) File
00 001 1402 203 1504 405 506 607 708 809 1610 1011 1112 1213 1314 1415 1516 16
166
APPENDIX H
SIMULATION LOG FILES
1st Run Setting Log File Inputs:
Landscape: CETIBELI.LCPWeather: CETIBELI.WTRWinds: CETIBELI.WNDAdjustments: CETIBELI.ADJFuel Moistures: CETIBELI.FMSConversions: NONECustom Fuel Models: NONECoarse Woody Fuels: NONEBurning Period: NONEProject File: F:\GGIT-TEZ-last\farsite\1\input\cetibeli.FPJBookmark File: F:\GGIT-TEZ-last\farsite\1\input\cetibeli1.BMK
Outputs:Shapefile: cetibeli.SHPRaster file: Raster File: cetibeli.toaRaster file: Raster File: cetibeli.fliRaster file: Raster File: cetibeli.rosRaster file: Raster File: cetibeli.fmlRaster file: Raster File: cetibeli.hpaRaster file: Raster File: cetibeli.sdrDisplay Units: METRICOutput File Units: METRIC
Model:Parameters: TimeStep 30.0Parameters: Visibles 30.0, 60.0Parameters: Perim Res 30.0Parameters: Dist Res 30.0Options: Crown Fire: DISABLEDOptions: Spotting: DISABLEDOptions: Spot Growth: DISABLEDOptions: Ignition Frequency: 5.0 %Options: Ignition Delay: 0 minsOptions: Fire Level Dist. CheckAcceleration: ONAcceleration: DEFAULTS
167
Post Frontal: OFFDead Fuel Moisture: PRE-CALCULATED
Simulate:Duration: Conditioning (Mo/Day): 08/15Duration: Starting (Mo/Day Hour:Min): 08/15 18:23Duration: Ending (Mo/Day Hour:Min): 08/16 16:45Options: Duration Reset: FALSEOptions: Restore Ignitions: FALSEOptions: Rotation Sensitive Ignitions: FALSEOptions: Show Fires as Grown: TRUEOptions: Ignition Spread Rates: FALSEOptions: Preserve Inactive Enclaves: TRUEOptions: Simulation Threads: 01
Attack:Ground Resources: NONEAir Resources: NONE
View:Viewport: MAXIMIZED
Output Log After Run Log File: F:\GGIT-TEZ-last\farsite\1\output\cetibeli.LGSDate File Created: 08\16\2003Time File Created: 02:58
Landscape File: F:\GGIT-TEZ-last\farsite\1\input\CETIBELI.LCPWeather File 1: CETIBELI.WTRWind File 1: CETIBELI.WNDAdjustment File: CETIBELI.ADJFuel Moisture File: CETIBELI.FMSConversion File: NoneCustom Fuel Model File: NoneCrown Fire: DisabledEmber Generation: DisabledBacking Spread: Calculated from Elliptical DimensionsAcceleration File Used: Default Values
Simulation Started (Day Hour:Min): 8/15 18:00Simulation Ended (Day Hour:Min): 08/16 16:23Elapsed Time (Days Hours:Mins): 00 22:00
Actual Time Step (min): 30.000000Visible Time Step (min): 30.000000Perimeter Resolution (m): 30.000000Distance Resolution (m): 30.000000
2nd Run Setting Log File Inputs:
Landscape: CETIBELI.LCPWeather: CETIBELI.WTRWinds: CETIBELI.WNDAdjustments: CETIBELI.ADJFuel Moistures: CETIBELI.FMSConversions: NONE
168
Custom Fuel Models: NONECoarse Woody Fuels: NONEBurning Period: NONEProject File: F:\GGIT-TEZ-last\farsite\2-
crown\input\cetibeli.FPJBookmark File: F:\GGIT-TEZ-last\farsite\2-
crown\input\cetibeli2.BMKOutputs:
Shapefile: cetibeli2.SHPRaster file: Raster File: cetibeli2.toaRaster file: Raster File: cetibeli2.fliRaster file: Raster File: cetibeli2.rosRaster file: Raster File: cetibeli2.fmlRaster file: Raster File: cetibeli2.hpaRaster file: Raster File: cetibeli2.cfrRaster file: Raster File: cetibeli2.sdrDisplay Units: METRICOutput File Units: METRIC
Model:Parameters: TimeStep 30.0Parameters: Visibles 30.0, 60.0Parameters: Perim Res 30.0Parameters: Dist Res 30.0Options: Crown Fire: ENABLEDOptions: Spotting: DISABLEDOptions: Spot Growth: DISABLEDOptions: Ignition Frequency: 5.0 %Options: Ignition Delay: 0 minsOptions: Fire Level Dist. CheckAcceleration: ONAcceleration: DEFAULTSPost Frontal: OFFDead Fuel Moisture: PRE-CALCULATED
Simulate:Duration: Conditioning (Mo/Day): 08/15Duration: Starting (Mo/Day Hour:Min): 08/15 18:23Duration: Ending (Mo/Day Hour:Min): 08/16 16:45Options: Duration Reset: FALSEOptions: Restore Ignitions: FALSEOptions: Rotation Sensitive Ignitions: FALSEOptions: Show Fires as Grown: TRUEOptions: Ignition Spread Rates: FALSEOptions: Preserve Inactive Enclaves: TRUEOptions: Simulation Threads: 01
Attack:Ground Resources: NONEAir Resources: NONE
View:Viewport: MAXIMIZED
Output Log After Run Log File: F:\GGIT-TEZ-last\farsite\2-crown\output\cetibeli2.LGSDate File Created: 08\18\2003Time File Created: 02:40
Landscape File: F:\GGIT-TEZ-last\farsite\2-crown\input\CETIBELI.LCP
169
Weather File 1: CETIBELI.WTRWind File 1: CETIBELI.WNDAdjustment File: CETIBELI.ADJFuel Moisture File: CETIBELI.FMSConversion File: NoneCustom Fuel Model File: NoneCrown Fire: EnabledCrown Density LINKED to Crown CoverEmber Generation: DisabledBacking Spread: Calculated from Elliptical DimensionsAcceleration File Used: Default Values
Simulation Started (Day Hour:Min): 8/15 18:00Simulation Ended (Day Hour:Min): 08/16 16:23Elapsed Time (Days Hours:Mins): 00 22:00
Actual Time Step (min): 30.000000Visible Time Step (min): 30.000000Perimeter Resolution (m): 30.000000Distance Resolution (m): 30.000000
Final Run Inputs:
Landscape: CETIBELI.LCPWeather: CETIBELI.WTRWinds: CETIBELI.WNDAdjustments: CETIBELI.ADJFuel Moistures: CETIBELI.FMSConversions: CETIBELI.CNVCustom Fuel Models: CETIBELI.FMDCoarse Woody Fuels: NONEBurning Period: NONEProject File: F:\GGIT-TEZ-last\farsite\3-adjust\input-
1\cetibeli.FPJBookmark File: F:\GGIT-TEZ-last\farsite\3-adjust\input-
1\cetibeli3-1.BMKOutputs:
Shapefile: cetibeli.SHPRaster file: Raster File: cetibeli.toaRaster file: Raster File: cetibeli.fliRaster file: Raster File: cetibeli.rosRaster file: Raster File: cetibeli.fmlRaster file: Raster File: cetibeli.hpaRaster file: Raster File: cetibeli.cfrRaster file: Raster File: cetibeli.sdrDisplay Units: METRICOutput File Units: METRIC
Model:Parameters: TimeStep 30.0Parameters: Visibles 30.0, 60.0Parameters: Perim Res 30.0Parameters: Dist Res 30.0Options: Crown Fire: ENABLEDOptions: Spotting: DISABLEDOptions: Spot Growth: DISABLEDOptions: Ignition Frequency: 5.0 %Options: Ignition Delay: 0 mins
170
Options: Fire Level Dist. CheckAcceleration: ONAcceleration: F:\GGIT-TEZ-last\farsite\3-
adjust\input\cetibeli.ACLPost Frontal: OFFDead Fuel Moisture: PRE-CALCULATED
Simulate:Duration: Conditioning (Mo/Day): 08/15Duration: Starting (Mo/Day Hour:Min): 08/15 18:23Duration: Ending (Mo/Day Hour:Min): 08/16 16:45Options: Duration Reset: FALSEOptions: Restore Ignitions: FALSEOptions: Rotation Sensitive Ignitions: FALSEOptions: Show Fires as Grown: TRUEOptions: Ignition Spread Rates: FALSEOptions: Preserve Inactive Enclaves: TRUEOptions: Simulation Threads: 01
Attack:Ground Resources: NONEAir Resources: NONE
View:Viewport: MAXIMIZED
171
APPENDIX I
FIRE OBSERVATION DATA SHEETS AND
DESCRIPTIONS
(Rothermel and Rinehart, 1983)
172
173
174
175